# QaribuBot > QaribuBot is an AI website chatbot with a one-time trial and paid packages: paste one snippet and it reads your site's content to answer visitor questions around the clock. QaribuBot is a product of Qaribuhub Solutions LTD, a software company in Nairobi, Kenya. Merchants register a website, paste a snippet that only works on that site, and QaribuBot reads public pages plus any FAQs the merchant adds so answers can draw on that content. ## Core product information - [Home](https://qaribubot.com/): Overview of QaribuBot — trial, automatic site reading, and one-snippet installation. - [How it works](https://qaribubot.com/how-it-works): The setup flow — register a website, paste the snippet, QaribuBot reads your content, visitors get answers. - [Pricing](https://qaribubot.com/pricing): Live trial terms and paid Starter, Pro, and Business packages loaded from the plan service. - [Use cases](https://qaribubot.com/use-cases): How QaribuBot fits different kinds of websites. - [Documentation](https://qaribubot.com/docs): Setup, knowledge, widget customization, conversations, team, and billing. - [Blog](https://qaribubot.com/blog): Guides on setup, pricing, best practices, privacy, and use cases for website AI chatbots. ## Use cases ### E-commerce chatbot Add a chatbot for your online store that reads your catalog and answers product, delivery, and return questions around the clock. - Reads product and policy pages automatically after you add your website URL - Answers sizing, stock, shipping, and return questions while shoppers are still on the page - Shows Call, Email, or WhatsApp buttons when those details are on your site - One installation snippet — no rebuild of the store you already run Source: https://qaribubot.com/use-cases/ecommerce-chatbot ### Real estate chatbot QaribuBot reads your listing pages so visitors get answers while you are on site visits. Conversations are saved so you can follow up. - Reads listings, neighbourhoods, and financing FAQs from your existing pages - Answers availability and service-charge questions from what you published - Works on WordPress, custom sites, and many property sites with one snippet - Review conversations in your dashboard and flag answers that need improvement Source: https://qaribubot.com/use-cases/real-estate-chatbot ### School & education chatbot Give parents and students a 24/7 answer to admissions, fee structure, and term-date questions from the pages you publish. - Reads admissions, fees, and calendar pages so answers follow the pages you keep current - Handles repeating parent questions when the answer is already on the site - Works for primary, secondary, and university sites with one snippet - Start with a free trial, then pick a plan that fits Source: https://qaribubot.com/use-cases/school-chatbot ### Business & company chatbot QaribuBot reads your services, pricing, and contact pages so visitors can get answers even when no one is at the desk. - Reads your services, about, and pricing pages automatically — no FAQ writing required to start - Answers service-area and quote questions from what you published - Review visitor questions in your dashboard so you can follow up - One snippet — no website rebuild; most sites only need someone who can paste it Source: https://qaribubot.com/use-cases/business-chatbot ### Government & public-service chatbot Let QaribuBot answer procedural questions about services, requirements, and office hours from your ministry, county, or agency site. - Reads service pages, requirements, and forms people already search for - Answers document and office-location questions from published pages - Handles routine procedural questions so staff can focus on cases that need a person - Deploys on existing government sites with one snippet, no system migration Source: https://qaribubot.com/use-cases/government-chatbot ### Healthcare & clinic chatbot QaribuBot answers general questions about services, hours, and insurance from your clinic or hospital site — it does not replace medical advice. - Reads your services, specialties, and insurance-accepted pages - Answers Saturday hours and insurance questions from those pages - Directs urgent or clinical questions to a phone number or booking page instead of guessing - One snippet install — sits on the same site as your existing booking system Source: https://qaribubot.com/use-cases/healthcare-chatbot ### Portfolio chatbot QaribuBot reads your portfolio and project pages so freelancers and studios can answer client questions automatically, day or night. - Reads your projects, case studies, and rates pages - Answers branding, turnaround, and availability questions from those pages - Review interested visitors' questions in your dashboard while you are on client work - One snippet — works on Webflow, WordPress, or a custom-built site Source: https://qaribubot.com/use-cases/portfolio-chatbot ### News site chatbot QaribuBot indexes your published articles so readers can ask what a story covers instead of scanning headlines. - Reads published articles and category pages automatically after you add the site - Points readers to the right story when that coverage was crawled - Keeps up to date through the refresh schedule on your plan - One snippet — sits alongside your existing website builder Source: https://qaribubot.com/use-cases/news-chatbot ### Blog chatbot QaribuBot reads your posts so readers can ask a question and get pointed straight to the relevant article, instead of searching manually. - Reads your posts (within your plan’s page allowance) so answers draw on your archive - Points readers to a matching post when it was read - Keeps pace with new posts through your plan's configured refresh schedule - One snippet — works on typical blogging platforms that allow a snippet Source: https://qaribubot.com/use-cases/blog-chatbot ### NGO & non-profit chatbot QaribuBot reads your programs, eligibility criteria, and how-to-donate pages so routine questions can be answered from that content. - Reads your programs, eligibility, and donation pages automatically after you add the site - Answers apply and donate questions from those pages - Handles routine questions so staff can spend more time on fieldwork - One snippet — start with the trial, then pick a plan that fits Source: https://qaribubot.com/use-cases/ngo-chatbot ### Custom AI chatbot for any website Whatever your site is about, QaribuBot reads it and answers visitor questions from your actual content — no industry template required. - Reads whatever pages you publish, no predefined category needed - Answers in the language a visitor uses, matched to your published content - Same one-snippet install regardless of your site's platform - Start a free trial and upgrade only if you need more chat volume or websites Source: https://qaribubot.com/use-cases/custom-chatbot ## Getting started ### Create your account Sign up, verify your email, and land in your dashboard. Click **Start free trial** from the Pricing page, or **Log in** if you already have an account. Create an account with email and password, or use **Continue with Google**. Google sign-in skips the email code step when Google has already verified that address. If you signed up with email, we send a 6-digit verification code. Enter it on **Verify your email** and click **Verify & continue**. Didn’t get it? use **Resend code** — there is a short wait between sends. After verification you land in your dashboard, starting on **Billing** so you can start the trial or pick a paid plan. We also send a welcome email with the first setup steps: connect a website, paste the snippet from **Domains**, then open **Widget**. From the sidebar you can open **Dashboard**, **Domains**, **Conversations**, **Widget**, **Knowledge**, **Team**, and **Billing**. Your name in the sidebar opens **Profile** (display name, company, photo, password, a copy of your data, and account deletion). Need the public walkthrough instead of the dashboard? Stay on this Docs site — it follows the product as it actually works, not a generic chatbot guide. Source: https://qaribubot.com/docs/create-your-account ### Starting a free trial vs. picking a plan What a free trial requires, what each plan unlocks, and how it converts. Each account can start **one** free trial. Current length, chat allowance, and website limit are shown on the live plan cards on the Pricing page and again at checkout — we do not freeze those numbers in this article because they can change. The trial uses a card: you choose the paid plan you want to continue on, add a card, and get full access to that plan for the trial window. Near the end we email a trial-ending reminder. If you do not cancel, it converts automatically to a paid subscription at that plan’s normal price. If this account already used its trial, the public pricing grid hides the trial card and Billing takes you straight to paid checkout. Paid plans (Starter, Pro, Business) differ by monthly chat cap, how many websites you can connect, how often we re-read your pages, team seats, and whether you can export transcripts. Open **Billing** or the Pricing page for the live numbers. All current paid plans include conversation history and widget customization; transcript export starts on Pro; extra team seats start on Business. Prefer not to use a card? **M-Pesa** is available as prepaid access for Kenya on the Billing page. Access lasts for the period you pay, then you top up again from Billing. The **Dashboard** sidebar shows chats used vs the cap for the current billing period. When the cap is reached, new visitor chats show an error until the period resets or you upgrade — your knowledge and snippet stay in place. Source: https://qaribubot.com/docs/trial-and-plans ### What’s on the Dashboard Usage, install status, knowledge, and recent chats in one place. After you connect a website, **Dashboard** is a live snapshot for the website currently selected in the website switcher. You will see: current plan name, **requests this period** (used / cap), number of connected websites, how many knowledge chunks you have, conversation count, and whether the chatbot is live on your site. The usage chart plots chats over time. You can change the range from that chart. **Recent conversations** links into the full thread. If no website is connected yet, Dashboard asks you to add one first — nothing else in the product can run without a registered website. Source: https://qaribubot.com/docs/dashboard-overview ## Connect your website ### Register your website Add a website from the Domains page before anything else. Go to **Domains** and add the website address visitors actually use (for example `www.yourshop.com` or `yourshop.com`). Each plan has a connected-website limit. If adding another website is blocked, the page names the plan that includes more sites. Each website is separate: its own key, widget settings, knowledge, conversations, and a badge that shows whether the chatbot is live. Use the website switcher when you have more than one. Registering a website automatically starts the first read of your pages. You do not need to press **Start crawl** that first time. `www` and the same address without `www` are treated as the same website. A different subdomain is a different site (`shop.example.com` is not `example.com`) and needs its own registration if you want the chatbot there. You can remove a website from this page. That deletes that site’s key, widget settings, and knowledge. Confirm before you delete. If you outgrow the site limit, upgrade on **Billing** rather than pasting one snippet on two unrelated websites — the chatbot only answers on the website you registered. Source: https://qaribubot.com/docs/register-a-domain ### Install the chatbot widget Where to find the snippet, how to embed it on your website, and where it goes. On **Domains**, click ** Code snippet** next to the website. That dialog is the source of truth for the live snippet. If the full key is not visible yet, click **Refresh key** once first. Refreshing **stops the previous key immediately**, so update every page that still has the old snippet. The snippet looks like this (your copy will differ — paste the one from the dialog, not this example): `` Copy it exactly as shown. Do not change the addresses or the key. The snippet only works on the website you registered. Paste it at the end of every page that should show the chatbot, just before the page finishes. One snippet in a shared footer or theme layout is enough if that layout wraps the whole site. There is a **Copy snippet** button in the dialog, and **Copy key** if you only need the key. The chatbot is a button in the bottom-right or bottom-left corner (you choose on **Widget**). Visitors do not need an account. Chats are stored on your Conversations page. After it is live, set the greeting they see first — see [Customize how your chatbot looks and talks](/docs/customize-appearance). Source: https://qaribubot.com/docs/install-the-widget ### Add the chatbot to WordPress, Shopify, Wix, and Webflow Where to embed the chatbot widget in common site builders. QaribuBot is one snippet. You do not install a separate WordPress or Shopify app from us — you paste the snippet from **Domains** into whatever place your platform uses for site-wide footer code. - **WordPress (theme):** Appearance → Theme File Editor → your theme footer (often `footer.php`), at the end of the page. Prefer a child theme so updates do not remove it. A header/footer or “insert scripts” plugin that adds code at the end of the page is safer if you do not edit theme files. - **WordPress.com / hosted builders:** use the custom code / footer injection setting if your plan allows it. - **Shopify:** Online Store → Themes → … → Edit code → `layout/theme.liquid`, at the end of the page. - **Webflow:** Project settings → Custom code → Footer code. - **Squarespace:** Settings → Advanced → Code injection → Footer. - **Wix:** Settings → Custom code → Add code to body end, site-wide. - **Custom-built sites:** put the snippet in the main page template so it loads on every page. - **Google Tag Manager:** Custom HTML tag, All Pages, with the exact snippet. Publish the container. Some ad blockers still hide Tag Manager; if the chatbot never shows as live, paste the snippet into the theme instead. After you paste it once, the chatbot stays as visitors click around your site. It still needs to be present on the first page they load. If the live website address differs from the one on **Domains** (a test site vs your real site, or a new custom address), register that address or the chatbot will not answer there. If the widget still does not appear, see [Confirm the widget is live](/docs/verify-install) and [Widget not showing up on my site](/docs/widget-not-showing). Source: https://qaribubot.com/docs/install-on-platforms ### Confirm the widget is live How we detect the chatbot on your site and what the badge means. Once the snippet is live and someone opens the page, we mark the website **Widget install verified ✓** on **Domains**, **Widget**, and **Dashboard**. There is nothing to click to confirm it manually. Until then you will see **Not detected on your site yet**. If it stays unverified: confirm the snippet is on the page you are actually opening (view source and search for the snippet), the website address matches **Domains** (`www` is fine), and you are not testing with an ad blocker or privacy extension that blocks the chatbot. Those extensions can make a working install look broken on *your* machine while real visitors still chat. Open the page in a private window with extensions disabled, or from a phone on cellular, then refresh **Domains**. Source: https://qaribubot.com/docs/verify-install ### Your key only works on your website Why the chatbot only answers from the registered website. Every website has one live key. The chatbot only answers when that key is used on the website you registered (`www` is fine). That is intentional: someone who copies your snippet onto another site will not get answers from your pages. **Refresh key** on **Domains** issues a new key and the old one stops working immediately. Update the snippet everywhere (theme, Tag Manager, test site, live site). There is no overlap window. If the chatbot refuses to load, the page address is probably not the one you registered. Add the missing address as another website (if your plan allows) or make sure visitors use the same address you registered. Opening a saved page from your computer will not work. The chatbot needs to run on your live website. Source: https://qaribubot.com/docs/origin-and-api-key ## Crawling & knowledge ### How we learn your pages Automatic schedule, what we read, manual runs, and statuses. The chatbot answers from **your** pages and from text you add yourself. We do not teach other customers your content; answers use only what is saved for that website. We start reading your pages automatically when you first register a website. After that, re-reads are a **plan credit**: trial includes that first read only; Starter is 1 per month; Pro is 1 per week; Business is 1 per day. **Start crawl for [your website]** uses the same credit — if the automatic read already ran this period, the button is locked with an upgrade prompt. Because a scheduled re-read would otherwise spend the only credit, you can **turn off automatic recrawls** on **Knowledge**. That keeps the period’s credit for a manual run. The page shows the next credit or next automatic read in your local timezone. A completed read **replaces previously saved pages** for that website. The **latest read** is what the bot prefers when facts disagree — contact details, hours, prices, product names, or anything else that changed on the live site. Answers you typed yourself remain, but if they conflict with the newest read, the live site wins. Failed or empty reads leave the previous answers in place. We stay on your website. We follow internal links, read sitemaps when your plan allows, and can pick up product, category, and blog pages. We skip files such as PDFs and images. Other websites’ pages are not read. Pages that load extra content after they open are still read, so prices, FAQs, and catalog details that appear after load can still be learned. We also pick up contact details, hours, and listed prices when they appear on the page. From that pass we save the page text plus contact facts when we find them: phones, emails, WhatsApp links, physical addresses, hours, and listed prices. Those facts are what the bot quotes for contact, hours, location, and listed prices. While a read runs you will see **Queued**, **Processing** (with page counts), **Completed**, **Failed**, or **Canceled**. If it fails, the reason is under the progress bar — not a bare “failed”. How much we can read (pages and time) is capped by your plan so a huge site on a small plan will not ingest everything. If answers miss a deep page, upgrade or add that content yourself, then read again. Source: https://qaribubot.com/docs/how-crawling-works ### Add answers yourself FAQs, policies, and anything we should not have to guess. On **Knowledge**, use **Add knowledge manually** for FAQs, return policies, shipping rules, or phrasing you want word-for-word. Title defaults to “FAQ”; change it to whatever you will recognise later. Text must be at least **20 characters** of real content. Empty or tiny snippets are rejected. Each entry can be opened, edited, or deleted on its own. Answers you typed yourself survive re-reads; pages we read are refreshed when a read completes. Use this for answers that are not on a public page, or for a page we cannot reach (login walls, PDFs). If it *is* on a public page, prefer a crawl so the bot also gets the link to share. The list is paged. Saved-answer counts also appear on **Dashboard**. Source: https://qaribubot.com/docs/add-knowledge-manually ### How the chatbot answers visitors Staff voice, knowledge, contact buttons, links, and what it will not invent. The widget is staff for **that website**. It speaks as “I” / “we”, not as a third-party narrator. Tone, greeting, name, and a short persona (up to 240 characters) are set on **Widget** — full steps in [Customize how your chatbot looks and talks](/docs/customize-appearance). Answers are meant to come from pages we read, contact facts, and answers you added. It is instructed not to invent prices, policies, or features that are not in that material — still review chats, because AI can get a detail wrong. If a visitor asks for a page that exists in your knowledge (blog, pricing, docs, a named article), the bot pastes the real web address. It will not guess a page such as a contact page if that page was never read. Phone, email, and WhatsApp are **not** pasted as text. When those details were found on the site, the widget shows **Call / Email / WhatsApp** buttons instead. Ask “how do I contact you?” to see them. Comparison questions (“is there something better?”) are answered confidently about **this** business. The bot should not recommend other vendors or shrug about “other solutions”. Chats keep a short memory of the current visitor thread (same browser session) so follow-ups stay on topic. Visitors can thumbs-up or thumbs-down a reply; you see those votes on **Conversations**, and admins see platform totals on **Chat quality**. Unsafe requests are refused in one line, then the bot offers website help again. If a listed fact is genuinely missing, it says it is not sure and offers the contact buttons — it should not invent an answer to fill the gap. Source: https://qaribubot.com/docs/how-the-bot-answers ## Widget customization ### Customize how your chatbot looks and talks Customize your website chatbot: name, greeting, three colours, Day or Night, position, tone, reply length, and persona. Choose three colours for your chatbot, then select a light or dark background. The preview updates as you make changes, and your saved settings are applied to the chatbot on your website. Open **Widget**. Settings apply to the website currently selected. If no website is connected, connect one first. **Widget customization** controls how the chatbot looks: - **Bot name** (up to 40 characters) — shown in the header. - **Greeting text** (up to 200 characters) — this is the first message visitors see when they open the widget on your website. Save, then refresh a live page to confirm it. - **Primary colour** — launcher button, header, and send button. Use the colour picker or type a 6-digit hex code. Restore default returns to QaribuBot gold (#F9C904). - **Secondary colour** — visitor message bubbles and supporting buttons such as Call. Restore default returns to QaribuBot teal (#54B7B6). - **Accent colour** — links, the focus ring around the message box, and small highlights. Restore default returns to QaribuBot teal (#54B7B6). - **Appearance** — **Day** (white chat background) or **Night** (black chat background). This is your choice for this website; it does not follow the visitor’s device theme. - **Position** — bottom right or bottom left. The widget picks light or dark text on each colour so labels stay readable. If a colour would make links hard to read, you will see a warning suggesting a darker or lighter accent. **Widget settings** control how the chatbot talks: - **Tone** — Friendly, Professional, or Formal. This changes how answers are phrased, not the facts. - **Reply length** — Short (about 1–2 sentences) or Medium (about 2–4). - **Persona / instructions** (up to 240 characters) — extra voice notes, e.g. “We are a clinic in Westlands; be calm and clear.” Instructions that try to override safety are ignored. The **Live preview** updates as you type, before you save — including colours and Day or Night. Click **Save widget**. Changes go live on your site within a few seconds; visitors do not need to keep refreshing forever, though a normal refresh picks up the new look immediately. If widget customization is not on your plan, the form is locked and the page names the plan that includes it. Current paid plans include this; a lapsed or unpaid account does not. Install status is repeated at the bottom of this page so you can see whether the snippet has checked in. Source: https://qaribubot.com/docs/customize-appearance ## Conversations ### View conversations and flag weak answers Browse chats, search, filter, and mark ones that need improvement. **Conversations** lists every chat for the selected website. Open a row to read the full thread. Search filters the list by preview text. **Needs better answer** shows only flagged threads. Check **Needs a better answer** on a conversation when the bot missed something. Use that list as a punch-list: add a Knowledge entry or fix the page, then recrawl. If conversation history is not on the current plan, the page is locked and names the plan that includes it. While a trial or paid plan is active, history is included on current plans. Threads are per website. Switch website in the header to see another site’s chats. Source: https://qaribubot.com/docs/view-and-flag ### Export a transcript CSV or JSON export on plans that include transcript export. If your plan includes transcript export (Pro and Business on current pricing), open a conversation and use **Export CSV** or **Export JSON** in the thread toolbar. If those buttons are missing or locked, your current plan does not include transcript export — check **Billing** or the Pricing page. Exports are for the open thread, not the whole account. For a full copy of your data (websites, knowledge, conversations), use **Download my data** on **Profile**. Source: https://qaribubot.com/docs/export-transcripts ### Request limits and what happens at the cap How visitor messages count, where to watch usage, and how to continue. Each visitor message that we answer counts as one request against the account’s monthly cap (shared across your websites). The cap is on the plan card and on **Dashboard** / **Billing** as used / remaining. When the cap is reached, new chats cannot be answered until the billing period resets or you upgrade. The chatbot stays visible; the visitor sees an error that the limit was reached. Usage is account-wide, not per website. Two busy websites on one Pro account share one pool. If you are hitting the cap early, upgrade on **Billing**, or reduce duplicate test chats while you are installing. Source: https://qaribubot.com/docs/request-limits ## Team ### Invite teammates Owner, admin, and member roles, seats, and invites. The person who created the account is the **owner** and always occupies a seat. They cannot be removed or demoted from **Team**. On plans with extra seats (Business on current pricing), **Team** lets an owner or admin enter an email, choose **Admin** or **Member**, and send an invite. Pending invites show an expiry and can be **Revoked**. Accepted members can have their role changed or be **Removed**. - **Admin** — connect websites, customize the chatbot, manage knowledge and crawls, invite or remove other admins/members (not the owner). - **Member** — view the dashboard and conversations, and flag threads that need a better answer. Billing, profile, data download, and account deletion stay with the **owner**. The invitee must sign in with the **same verified email** the invite was sent to, then open the invite link (or **Team** accept screen) to join. Seat math: owner + active members + pending invites cannot exceed the plan’s team seat limit. If inviting is blocked, the page says the current plan includes the owner only — upgrade on **Billing**. On Starter and Pro, the Team page explains that inviting teammates is not included. Source: https://qaribubot.com/docs/invite-teammates ## Billing & plans ### Card vs M-Pesa The two ways to pay, and how they differ. Open **Billing**. Choose a plan, then a payment method. **Card & other payment methods** is the recurring subscription: Visa, Mastercard, Amex, PayPal, Apple Pay, Google Pay, and similar checkout options. It auto-renews each billing period and supports self-service cancel/resume and **Manage billing** for invoices and cards. **M-Pesa** is prepaid for the period you choose, in Kenya. Pay from your phone at checkout. Access runs until that period ends; then top up again from Billing. Plan changes on M-Pesa happen the next time you prepay for the new plan. Live prices, chat caps, site limits, and trial length are always on the Pricing page and on the Billing plan list — use those, not a screenshot from last month. Source: https://qaribubot.com/docs/payment-methods ### Change or upgrade your plan Switching plans on an active card subscription or M-Pesa period. On an active card subscription, pick another plan on **Billing** and use **Upgrade to [plan]**. Billing prorates on the card subscription. If a lower plan would drop you under what you already use (too many websites, too many seats, or chats already above the smaller cap), Billing warns you before you confirm. On M-Pesa, changing plans means prepaying for the new plan’s period at the next checkout rather than an in-place proration. Website, crawl, and team limits change with the plan. Extra websites beyond the new limit need to be removed before a downgrade can succeed. Source: https://qaribubot.com/docs/change-plan ### Cancel or resume a subscription What happens to the chatbot when you cancel. Canceling a paid card subscription keeps access until the end of the current billing period. After that there is no active plan until you resubscribe — the snippet may still load, but chats will not be served. Canceling during a free trial is different: you keep access until the trial ends, the card is not charged, and afterward the account has no active plan. Before the period actually ends, use **Resume subscription** (or **Resume trial conversion** during a trial) to undo the scheduled cancellation. **Manage billing** updates invoices and payment methods on the card subscription. M-Pesa access simply ends when the prepaid window ends unless you top up. Source: https://qaribubot.com/docs/cancel-and-resume ## Account & data ### Profile, export, and delete your account Name, photo, password, download a copy of your data, or delete the account. Open **Profile** from your name in the dashboard sidebar. You can set display name, company name, and a profile photo (PNG, JPEG, or WEBP, under 400KB). Email notification and weekly digest checkboxes are saved with the profile. Password changes require at least 6 characters (email/password accounts). Google-only accounts manage the password at Google. **Download all my data** (owner only) saves a copy of your account: profile and photo, websites, widget settings, knowledge, conversations, team, your plan, usage, and payment amounts. Keys and other internal details are omitted. Use this before you request deletion if you want a backup. **Delete my account** is also owner-only and is **not immediate**. Choose one of the listed reasons (or Other, with a short note), type `DELETE`, and confirm. We email you that the request was received and will complete in **3 days**. That email includes **Cancel deletion**. On Profile, the delete button becomes **Deletion in {x} hours, press here to cancel**. Cancelling sends a confirmation email and leaves the account fully active. When the 3 days end, we **permanently delete** the account: knowledge, conversations, websites, widget settings, team access, your login, and your profile photo. Card subscriptions are cancelled immediately when the account is deleted. Payment amounts stay in our books without your name so income totals stay accurate. Payment processors may still hold their own receipts as required by law. Legal detail is in the Privacy Policy (how to download and delete) and the Terms (the 3-day window). Those pages are the contract; this article is the dashboard how-to. Source: https://qaribubot.com/docs/profile-and-data ## Troubleshooting ### Widget not showing up on my site Checklist for a missing, blocked, or unverified launcher. Confirm the snippet is at the end of the page the browser actually loaded (View Source, search for the snippet). A template you saved but did not publish will not count. Confirm the website address matches **Domains**. `www` vs the same address without `www` is fine. A preview URL, a builder subdomain, or a shop platform address vs your custom domain is **not** the same site unless you registered that address too. If you clicked **Refresh key**, every copy of the snippet must use the new key. The old key fails immediately. Ad blockers, Brave shields, and some privacy extensions block the request. Test in a private window with extensions off, or on a phone without those extensions. If your website uses extra security settings that block outside code, ask whoever manages the site to allow QaribuBot. If those settings omit us, the chatbot never appears. Theme styling that hides overflowing content can clip the launcher. If a cookie banner covers it, adjust that banner, not the snippet. Still stuck? Open the page in a private window, confirm the snippet matches **Domains**, then contact support if it still does not appear. Source: https://qaribubot.com/docs/widget-not-showing ### My crawl failed Reading the failure reason and what to do next. Open **Knowledge**. If the latest read failed, the reason is under the progress bar. Typical causes: the site was unreachable, the website address does not serve a public page, or your plan changed while the read was running. Fix the site, then **Start crawl** again. Login-only pages and calendars that never end will yield thin answers. Add the important answers yourself, or publish a public FAQ page and recrawl. Very large sites on a small plan hit page/time caps and look “done” without every URL. That is a plan limit, not necessarily a failure — upgrade or add key pages yourself. You can add knowledge manually at any time; it does not depend on the crawler. Source: https://qaribubot.com/docs/crawl-failed ### The bot is wrong, vague, or hedging Fix knowledge, persona, and recrawl so answers match the site. The bot only knows what is in **Knowledge** for that website. If a page was not read, or the live page changed after the last read, the answer will be stale or missing. Fix the source page (or add a manual FAQ), then **Start crawl** if you still have a crawl credit this period. Visitors can thumbs-up or thumbs-down a reply in the widget; you will see that vote on the chat. Flag the thread with **Needs a better answer** as a reminder list. For voice: set **Persona** and **Tone** on **Widget**. Ask it to speak as staff if you want first person (“we deliver…”) rather than “the company delivers…”. For prices: make sure the live page shows the price in text or a normal product/plan table, then recrawl. Do not expect the bot to invent a number that is not on the site. For contact: it will offer buttons instead of pasting numbers. If buttons are missing, put a phone, email, or WhatsApp link on a public page and recrawl. If visitors hit the monthly request cap, they will see a limit error rather than a normal answer — check **Dashboard** usage. Source: https://qaribubot.com/docs/bot-wrong-answers ## Blog ### How to Add an AI Chatbot to Your Website The real steps behind adding a working AI chatbot to your site — from choosing a tool to your first live conversation. Most guides make chatbot setup sound like a development project, with diagrams of APIs and webhooks and training pipelines. In practice, for the overwhelming majority of small and mid-sized websites, adding an AI chatbot is closer to installing an analytics tag than building software. This guide walks through exactly what happens, step by step, so you know what to expect before you start — and so you can tell the difference between a platform built for people like you and one built for an enterprise IT department. #### 1. Pick a chatbot that reads your own content The single biggest factor in whether a chatbot is useful is whether it actually knows your business. Generic chatbots answer generic questions — they're trained on the internet at large, not on your delivery policy or your product catalog. Look for a tool that can crawl your site automatically, reading your product pages, FAQs, and policies, rather than one that requires you to write every answer by hand before it's useful. This distinction matters more than almost any other feature on a comparison chart. A chatbot that needs a manually built knowledge base is a content project before it's a chatbot project — you're writing a small encyclopedia about your business before a single visitor gets a useful answer. A chatbot that crawls your existing pages gets you to a working baseline immediately, and manual entries become the exception (filling small gaps) rather than the rule. #### 2. Register your domain Most modern chatbot platforms tie your account to specific websites so your chat allowance can't be used by someone else's site, and so the platform knows which content to read. This step usually takes under a minute — you type in your website address, the platform checks it, and you're given a key that only works on that website. If you run more than one website — a main site and a separate landing page for a campaign, say, or storefronts for different brands — check whether the platform lets one account cover multiple domains. Managing five separate subscriptions for five small sites is a real, avoidable cost most businesses don't need to carry. #### 3. Paste one snippet This is the part people expect to be hard and usually isn't. You copy a single line of code and paste it before the closing tag of your site, or drop it into your theme's custom-code area if you're on WordPress, Shopify, Wix, or Squarespace. There's no build step, no npm install, no server configuration, and nothing that requires a developer's local environment. QaribuBot's [installation guide](/docs/install-the-widget) shows exactly where the snippet lives in the dashboard. If your site already uses Google Tag Manager for other scripts — analytics, ad pixels, and so on — you can add the chatbot the same way, as one more tag, with zero direct changes to your site's codebase. This is often the least disruptive path if you have a marketing team managing tags separately from your developers. #### 4. Let it crawl your content Once installed, a good chatbot platform automatically indexes your pages so it can answer questions about your products, pricing, and policies without you typing anything in by hand. This usually takes a few minutes for a small site and scales up for larger catalogs — a five-page business site might be fully indexed before you've finished reading this section, while a 2,000-product store could take longer. QaribuBot does this after you register the website: it reads public pages, with no FAQ writing required to get started. Paste the snippet so visitors can open the chatbot. You can watch crawl status in your dashboard, and run another read when your content changes (using the credits on your plan). See [How we learn your pages](/docs/how-crawling-works) for what a read includes. #### 5. Test it like a visitor would Before calling it done, ask the bot the three or four questions your visitors ask most — delivery times, pricing, how to book, whatever's specific to you. Try a few edge cases too: ask something it genuinely shouldn't know, and see whether it says so honestly or invents a plausible-sounding wrong answer. A trustworthy chatbot admits uncertainty rather than guessing. If an answer is off, most platforms let you add a quick FAQ snippet to correct it without waiting for a re-crawl. This is usually the fastest fix for a specific wrong answer — faster than editing the source page and waiting for the crawler to catch up. #### 6. Set the tone and give visitors a way out Two small details make a disproportionate difference to how the chatbot feels. First, match its tone to your brand — formal, professional, or casual — so it doesn't read like a bolted-on generic tool. Second, make sure there's always a visible path to a human: an email address, a phone number, or a live handoff. A visitor stuck in a loop with no way out is worse off than one who never met a chatbot at all. In QaribuBot those settings live on Widget — [customize how your chatbot looks and talks](/docs/customize-appearance). #### How long this actually takes Add it up: registering a domain (a couple of minutes), pasting the script (a couple more), waiting for the crawl (a few minutes), and testing with real questions (fifteen minutes if you're thorough) — most sites are fully set up within half an hour, and technically live from the moment the script is pasted. The refinement — reading early conversations and patching gaps — is worth doing over the following week, but it doesn't block go-live. Source: https://qaribubot.com/blog/how-to-set-up-ai-chatbot-website ### Best AI Chatbots for Small Business Websites What separates a genuinely useful small-business chatbot from an overbuilt enterprise tool you'll never fully configure. Small business owners don't have a dev team to babysit a chatbot integration, and they don't have the budget for enterprise per-seat pricing built for call centers. Most 'best chatbot' roundups are written for a generic buyer that doesn't quite exist — someone with unlimited budget and a technical team on standby. The right tool for a five-person business looks meaningfully different from the right tool for a 500-person support org, and the differences show up long before you compare feature checklists. #### Setup time is the first filter If a platform's onboarding involves a sales call, an implementation timeline, or a 'contact us for a demo' wall before you can even try it, it's probably not built with a small business in mind. Look for tools you can try free and have live the same day, without talking to anyone. This isn't just about convenience — it's a signal about who the product is actually designed for. Products built for small businesses tend to be self-serve by default, because their whole growth model depends on someone signing up alone at 9pm and having it work without a call. Products that gate you behind a sales conversation are usually optimized for a different kind of customer entirely. #### It should learn your content, not need it typed in The tools that scale down well to a five-page business site are the ones that crawl your existing pages automatically rather than requiring you to build a knowledge base from scratch. That's the difference between an afternoon project and a multi-week one, and for a small business, that difference often determines whether the chatbot gets set up at all. Ask specifically, before signing up: does this read my website automatically? If the answer involves a training interview, a spreadsheet template, or a 'content specialist' onboarding call, that's a workflow built for a company with a marketing team to spare, not a shop owner running the business themselves. #### Pricing that matches small-business volume A small business website might get a few hundred chat conversations a month, not tens of thousands. Look for plans priced around request volume with a genuinely usable free tier, rather than flat enterprise pricing that assumes call-center scale from day one. Watch specifically for per-seat pricing (charging per team member who can access the dashboard) or mandatory onboarding fees — these are reasonable for a 200-person support team and a poor fit for a business where one or two people will ever log in. #### One account, multiple domains If you run more than one site — a main site and a landing page, or storefronts under different brands — a platform that lets one account cover multiple domains without separate subscriptions saves real money as you grow. This is easy to overlook when you're setting up your first site, and expensive to discover later when you're paying for three separate plans that could have been one. #### What to actually test before committing Try the product on a real page and ask it the questions your customers ask most. Check how it handles a question outside your content. Check whether the widget looks right on a phone, since most of your traffic is probably mobile. None of this takes more than twenty minutes, and it tells you more than any comparison table. - Can you try it free, right now, without a sales call? - Does it crawl your site automatically, or do you write every answer? - Is pricing based on chat volume, not per-seat headcount? - Can one account cover more than one domain if you need it to? - Does the widget look right on an actual phone, not just a browser simulator? Source: https://qaribubot.com/blog/best-ai-chatbots-small-business-websites ### How to Choose the Right AI Chatbot for Your Business A short framework for evaluating chatbot vendors before you commit — beyond the feature comparison table. Every chatbot vendor's feature page looks similar: 'AI-powered,' '24/7 support,' 'easy setup.' This is a decision checklist — not another 'best of' ranking. Run every vendor through the same practical checks; for a scored comparison of small-business tools, see Best AI Chatbots for Small Business Websites. #### What does it actually know? Ask specifically: does it read my website automatically, or do I have to write every answer? A chatbot that only knows what you manually feed it will always lag behind your real content — new products, updated policies, seasonal changes — because someone has to remember to update it by hand. The honest way to test this is to sign up, connect a real page, and see what the bot says about something on that page you didn't type in yourself. If it gets it right, the crawling actually works. If it can't answer basic questions from your own homepage, no amount of manual FAQ writing will fully compensate. #### How does it handle what it doesn't know? A chatbot that confidently makes up an answer when it's unsure is worse than no chatbot, because a visitor has no way to tell a fabricated answer from a real one. Test edge cases deliberately — ask it something genuinely outside your site's content and see whether it says so honestly, offers to connect you with a human, or fabricates a plausible-sounding answer anyway. This single test tells you more about a chatbot's reliability than a dozen feature comparisons. Vendors rarely advertise how their product handles uncertainty, because 'sometimes it makes things up' isn't a selling point — you have to check for yourself. #### Can you see what it's actually saying? You should be able to review real conversation transcripts, not just a satisfaction score or a dashboard of vague engagement metrics. This is how you catch a bad answer before ten more visitors get the same one, and it's the difference between a chatbot you actively manage and one you've quietly lost visibility into. If a vendor's dashboard only shows aggregate numbers — total chats, average response time — and hides the actual conversation content behind a higher tier or doesn't offer it at all, that's worth noting before you commit. #### What happens if you outgrow the plan — or want to leave? Check the upgrade path and whether there's lock-in. A widget that's one script tag to install is usually one script tag to remove, which is a reasonable sign the vendor isn't relying on switching costs to keep you rather than genuine usefulness. Ask what happens to your data — your conversation history, your crawled content — if you cancel. Also check whether upgrading actually requires a new conversation with sales, or whether it's a self-serve plan change. The latter is a good sign the company expects to earn your continued business rather than trap it. #### Putting it together None of these checks require technical expertise, and none of them take long — you can run this entire evaluation for a vendor in under an hour, most of it spent just using the free trial like a real visitor would. That's usually more revealing than any spec sheet the vendor publishes themselves. Source: https://qaribubot.com/blog/how-to-choose-right-ai-chatbot-for-business ### AI Chatbots for Websites: Complete Beginner's Guide Starting from zero: what a website AI chatbot actually is, how it works behind the scenes, and what it can realistically do for you. If you've never used one, 'AI chatbot' can sound like either magic or a gimmick, depending on who's selling it to you. It's neither. It's a fairly simple idea — a chat widget that reads your content and answers questions about it — wrapped in language-model technology that makes the conversation feel natural rather than robotic. This guide starts from zero and builds up to what you actually need to know to make a decision. #### What it is, in plain terms A website AI chatbot is a small chat window, usually in the bottom corner of your site, where visitors can type a question and get an answer generated from your actual content — not a script of pre-written replies to exact-match phrases, the way older-generation chatbots worked. If a visitor phrases a question in an unexpected way, a modern AI chatbot can still understand the intent and answer correctly, rather than failing because the wording didn't match a rule. This is the core upgrade over the chatbots people remember being frustrated by a decade ago. Those older tools worked on rigid decision trees — 'press 1 for billing' logic dressed up as chat. Language-model-based chatbots read and respond more like a person reading your FAQ page would. #### How it knows what to say Modern chatbots work by reading — crawling — your website's pages: products, FAQs, policies, about pages. That crawled content becomes the source material for answers. When a visitor asks something, the bot searches that material and generates a reply grounded in what your site actually says, rather than guessing from general knowledge about businesses like yours. This grounding step is why site-aware chatbots are more useful for business facts than a general assistant. A chatbot with no access to your site might invent a price. A chatbot that has read your pricing page can use your published numbers — still review replies, because AI can get details wrong. #### What it's good at Repetitive, factual questions: hours, pricing, delivery areas, how something works, whether you carry a product. Anything a human would answer the same way every single time is a good fit — which, for most businesses, turns out to be the majority of what visitors actually ask. It's also available around the clock in a way a person on the desk often isn't. The fiftieth visitor asking 'what are your hours' can get an answer from the same published hours as the first — including at 3am — without someone from your team being awake. Still review chats; the AI can get a detail wrong. #### Where a person still helps QaribuBot is strongest on factual questions from your own pages — pricing, hours, policies, product details. For complaints that need empathy, or anything that needs account access, the widget can point visitors to you with a call, email, or WhatsApp button. A good setup makes that handoff easy to find. See our guide on chatbot-to-human handoff for how that transition should actually work in practice. #### A realistic first-week expectation In your first week, expect the bot to answer most straightforward questions well, and to occasionally get something wrong or incomplete — usually because the relevant information isn't clearly written anywhere on your site yet, not because the AI is fundamentally unreliable. The fix is almost always the same: notice the gap, add a short FAQ entry or update the source page, and move on. This is completely normal and not a sign the tool isn't working. Source: https://qaribubot.com/blog/ai-chatbots-for-websites-beginners-guide ### Can You Add an AI Chatbot to Your Website Without Coding? You don't need to know how to code to add a working AI chatbot to your site. Here's what the actual process looks like. Short answer: yes, and it's one of the most common misconceptions keeping non-technical business owners from trying this. You don't need a developer, and you don't need to touch a line of code beyond copy-and-paste. This guide is the no-code path — for the full setup sequence, see [How to Add an AI Chatbot to Your Website](/blog/how-to-set-up-ai-chatbot-website). Platform-by-platform paste locations are in [How to Add an AI Chatbot to WordPress, Shopify, Wix, and Webflow](/blog/add-ai-chatbot-wordpress-shopify-wix-webflow) and the [docs platform guide](/docs/install-on-platforms). #### The 'no-code' part is a single snippet Most AI chatbot platforms, including QaribuBot, give you one snippet of code to paste into your site. This is called a script tag, and while it technically is code, using it requires no coding skill — the same way pasting a Google Analytics tracking code doesn't require you to understand JavaScript. On WordPress, that snippet goes into a plugin like 'Insert Headers and Footers' or your theme's custom-code box, both of which are point-and-click settings pages, not code editors in any meaningful sense. On Shopify, Wix, or Squarespace, there's usually a dedicated 'custom code' or 'embed' section built for exactly this purpose, sitting right in your normal site settings. #### You don't configure the AI itself You're not writing prompts, training a model, fine-tuning anything, or setting up a database. The platform crawls your site content automatically and handles the AI side entirely on the backend — your job is pasting the snippet and, optionally, reviewing how it's answering questions and adjusting a few settings like tone or color. This is a meaningfully different experience from, say, setting up an email automation workflow with branching logic, or configuring a CRM's custom fields. There's no flowchart to build and no logic to design — the crawling and the conversation handling are both automatic. #### Platform-by-platform, what it actually looks like On WordPress: install a header/footer plugin (free, widely used, five-minute setup), paste the snippet, save. On Shopify: go to your theme settings, find the custom code or 'additional scripts' section, paste, save. On Wix or Squarespace: use the built-in embed/custom-code widget from the site editor, drop it anywhere, publish. In every case, the total hands-on time is a few minutes, and none of it involves anything resembling programming — reading a settings page, pasting text, and clicking save. #### When you might still want technical help If your site is a fully custom build with an unusual content management setup — a bespoke framework, a legacy system with no plugin ecosystem — someone with basic HTML familiarity can paste the tag in under five minutes. That's a five-minute favor from anyone comfortable editing a template file, not a development project requiring a quote or a sprint. The only genuinely technical scenario is if your site has strict content security policies that block third-party scripts by default — in that case, whoever manages that configuration needs to allowlist the chatbot's domain, which is a one-line change, not a build. Source: https://qaribubot.com/blog/add-ai-chatbot-without-coding ### How to Add an AI Chatbot to WordPress, Shopify, Wix, and Webflow Where the same installation snippet goes on WordPress, Shopify, Wix, and Webflow — without installing a separate app. You do not need a separate WordPress plugin or Shopify app to add QaribuBot. After you register your website, you copy one installation snippet from Domains and paste it where your platform injects site-wide footer code. This article is the platform overview; the exact dashboard clicks are in [Add the chatbot to WordPress, Shopify, Wix, and Webflow](/docs/install-on-platforms). #### What you paste, on every platform The chatbot widget is one script tag. It is the same snippet on WordPress, Shopify, Wix, Webflow, and a custom site. Register the live website first so the key only works on that site, then paste the snippet so visitors can open the widget. After you add the website URL, QaribuBot starts reading public pages automatically. You do not need to write a knowledge base before the first answers. #### WordPress On self-hosted WordPress, paste the snippet in a header/footer plugin or your theme footer so it loads on every page. A child theme or an insert-scripts plugin is safer than editing a parent theme that updates. Hosted WordPress.com sites use the custom-code or footer injection setting if your plan allows it. You are not installing a QaribuBot plugin from the WordPress directory — you are embedding the chatbot widget with the snippet from your dashboard. #### Shopify On Shopify, open Online Store → Themes → Edit code and paste the snippet in layout/theme.liquid at the end of the page, or use your theme’s custom-code / additional scripts area if it has one. Publish the theme after you save. The chatbot reads the product and policy pages you already publish. It does not replace Shopify checkout or process orders. #### Wix On Wix, use Settings → Custom code → Add code to body end, site-wide. That is the same place most Wix sites use for analytics tags. Publish after you save so the widget appears on the live site, not only the editor preview. #### Webflow On Webflow, open Project settings → Custom code → Footer code, paste the snippet, and publish. If you use a reverse proxy or a custom domain, register the address visitors actually type. #### After you paste it Open a live page (not only a designer preview), check the widget on a phone, and ask a question that is clearly on your site. If it does not appear, confirm the snippet is on that page and that the website address matches Domains. For the exact copy from your dashboard, including Refresh key, see [Install the chatbot widget](/docs/install-the-widget) and [Add the chatbot to WordPress, Shopify, Wix, and Webflow](/docs/install-on-platforms). Source: https://qaribubot.com/blog/add-ai-chatbot-wordpress-shopify-wix-webflow ### How Much Does an AI Chatbot Cost for Small Websites? What AI chatbots actually cost, why pricing is usually based on chat volume, and how to avoid overpaying for a small site. Chatbot pricing pages can be confusing on purpose — request-based, seat-based, conversation-based, flat enterprise quotes hidden behind a 'contact sales' button. Here's what actually drives the cost, what a small site should realistically expect to pay, and how to avoid the two most common ways businesses overpay for a tool they only need lightly. #### Free tiers are usually real, with limits Trial and free-plan policies vary widely. Check whether access expires, whether a payment method is required, and whether the request and website allowances are large enough for a real evaluation. QaribuBot publishes its current trial terms directly from its plan configuration on the pricing page. The key question to ask about any 'free' plan is whether it's actually usable long-term or just a funnel into a forced upgrade. A free tier with a hard 14-day trial clock is a different product decision than a free tier with an ongoing monthly quota — the latter lets you genuinely test whether the tool earns its paid upgrade before you commit any money. #### What drives the price up The main lever is chat volume — how many conversations your widget handles per month — not features. A busier site simply needs a higher request cap, while implemented features such as transcript export, team seats, and crawl capacity may also differ by tier. QaribuBot publishes those current differences directly from its live plan configuration. This is a genuinely different pricing model from most SaaS tools, which price by seats or by feature tier. Chatbot pricing tracks usage because the underlying cost to the vendor scales with usage too — more conversations means more AI inference cost on their end, so it makes sense that your price scales the same way. #### Watch for per-seat or setup-fee pricing Some enterprise-oriented tools charge per team member who can log into the dashboard, or require a paid onboarding package before you can even test them. For a small business where one or two people will ever touch the dashboard, per-seat pricing adds cost with zero corresponding value — you're paying for a pricing model designed around a support team's org chart, not your actual usage. A genuine setup fee is a stronger red flag still. If a script-tag install genuinely takes minutes, a mandatory paid onboarding fee is pricing the sales process, not the product. #### A realistic small-site budget For a site getting a few hundred to a couple thousand chat conversations a month, entry-level chatbot packages are generally well below the cost of even part-time live-chat staffing. Compare current live prices against the request volume you actually expect. If your site is getting tens of thousands of conversations a month, you've likely outgrown 'small site' pricing tiers regardless of vendor, and it's worth comparing plans built specifically for that volume rather than assuming the entry-level tier scales indefinitely. #### How to estimate what you'll actually need A rough rule of thumb: expect somewhere between 1% and 5% of your monthly site visitors to open the chat widget and ask something, depending on how visible the widget is and how much visitors typically need help. If you get 5,000 monthly visitors, that's roughly 50–250 conversations a month — comfortably within most free or entry-level paid tiers. Source: https://qaribubot.com/blog/how-much-does-ai-chatbot-cost ### What Can an AI Chatbot Actually Do on Your Website? A grounded look at what a website AI chatbot is genuinely capable of — and how to pair it with a path to your team. A website chatbot is strongest when it answers from your real pages — pricing, hours, policies, products — instantly, at any hour. Here is what that looks like in practice, and how to pair it with a path to your team. #### What it does well Answering factual questions from your own content: pricing, hours, delivery areas, product specs, policies, how a process works. It does this instantly, at any hour, without a human on standby — and consistently, without a bad day or a distracted moment affecting the answer quality the way it might for a tired staff member on their fourth hour of the same question. It also handles volume gracefully in a way humans structurally can't. If fifty visitors ask the same question in the same ten minutes — during a sale, say, or after a policy change goes live — the chatbot answers all fifty instantly and identically, where a human team would create a queue. #### Capturing and organizing visitor questions Even when it can't fully resolve something, a chatbot logs what visitors are asking, which becomes useful data over time. You can see recurring questions you haven't addressed on your site yet — often revealing gaps in your own documentation you didn't know existed, because nobody had been tracking what people actually ask versus what you assumed they'd ask. #### When a person is the better next step For account-specific help, refunds, or a complaint that needs a real conversation, the chatbot can point visitors to you. Pair it with a visible call, email, or WhatsApp path so those visitors still get a clean way forward. That combination — instant answers on your published facts, plus an easy path to your team — is how most businesses get the most from a website chatbot. #### The realistic bar Think of it as a very fast, very patient front-desk person who's read your entire website and never gets tired of the same question — not a full customer service department, and not a decision-maker with authority to deviate from what's written down. That framing sets the right expectations for what to build around it, and what to keep routing to a human. #### Where the line moves as you customize it The exact boundary of 'what it can do' shifts a little depending on how much you invest in setup — adding detailed FAQ entries for edge cases, keeping your pages current, tuning tone for your brand. But the fundamental shape stays the same: it's strongest on repeatable, factual, low-stakes questions, and weakest on anything requiring judgment, empathy, or account-specific authority. Source: https://qaribubot.com/blog/what-can-ai-chatbot-do-on-website ### AI Chatbots vs. Live Chat: Which Is Better for Your Website? They solve overlapping but different problems — here's how to think about which one your site actually needs. This isn't really an either/or question for most businesses, but if you're choosing where to start — or trying to decide whether to keep paying for both — it helps to know precisely what each option is good at, rather than treating them as interchangeable versions of the same idea. #### Coverage Live chat is only as available as your team — usually business hours, sometimes with gaps during lunch, weekends, or busy periods when staff are stretched thin across other channels. An AI chatbot answers at 2am on a Sunday exactly as well as at 2pm on a Tuesday, with no scheduling, no time-zone coordination, and no coverage gaps to manage. For a business with international visitors or customers browsing outside normal hours — which is most e-commerce and increasingly most service businesses — this coverage gap is where a meaningful share of live-chat's potential value is simply never captured. #### Cost Staffing live chat, even part-time, costs more per month than most AI chatbot plans — and it scales with headcount, not just traffic, meaning growth in visitor volume directly increases your staffing cost. An AI chatbot's cost scales with chat volume instead, which tends to grow much more slowly than the cost of adding another live-chat agent. There's also a hidden cost to live chat that's easy to underweight: training time, turnover, and the inconsistency that comes from different agents answering the same question slightly differently. A chatbot answers with the same accuracy on day one as it does a year later. #### Where live chat still wins Genuinely complex, emotional, or high-stakes conversations — a frustrated customer with a real problem, a big B2B deal being negotiated, a nuanced technical question that depends on specifics not written anywhere — benefit from a human who can read the situation, ask clarifying questions naturally, and has real authority to fix things on the spot. There's also a trust dimension for certain purchases. For a high-value B2B sale, some buyers specifically want to know a real person is on the other end before committing, and no amount of AI polish substitutes for that in every context. #### The common setup Most businesses that use both let the AI chatbot handle the repetitive front line and escalate to a human — via a contact form, phone number, or live handoff — when a question needs a person. That combination covers more ground than either alone: the chatbot absorbs the volume of routine questions, freeing the human team's time for the conversations that actually benefit from a person's judgment. In practice live chat staffing may get more focused over time, not eliminated — the team can spend hours on conversations that need a person, instead of repeating the same published facts. Source: https://qaribubot.com/blog/ai-chatbot-vs-live-chat ### How to Train an AI Chatbot on Your Website Content 'Training' a modern website chatbot usually means something much simpler than it sounds. If you're picturing months of machine-learning work — labeling datasets, tuning models, running training jobs — that's not what this is for a website chatbot, and it's worth clearing that up before it scares anyone off. 'Training' here mostly means making sure the bot has the right source material, and then fixing the specific gaps that show up once real visitors start asking real questions. #### Start with the automatic crawl The bulk of the work should happen automatically — a good platform reads your existing pages and builds its knowledge from that, with no manual data entry required to get a working baseline. This is the single biggest labor-saver in the whole process: instead of transcribing your business knowledge into a chatbot-specific format, the platform reads what you've already written for your customers. This means your existing content quality directly determines your chatbot's quality on day one. A site with clear, complete product and policy pages produces a chatbot that answers well immediately. A site with thin or outdated content produces a chatbot that inherits those same gaps — which is itself useful information, since it often surfaces content problems you hadn't noticed. #### Fill gaps with FAQ snippets For questions that aren't clearly answered anywhere on your site — an unwritten policy, a common clarification you usually explain verbally or over email — most platforms let you add short FAQ entries the bot can pull from directly. This is the fastest way to fix a specific bad answer, faster than rewriting a page and waiting for a re-crawl. In QaribuBot that is [Add answers yourself](/docs/add-knowledge-manually). A good rule of thumb: if you've explained the same thing to three different customers by email or phone this month, it's worth a two-sentence FAQ entry. That single entry then answers it instantly for every future visitor who asks. #### Review real conversations The most useful 'training' step, ongoing, is reading actual transcripts periodically. Patterns jump out fast — the same misunderstood question, a policy visitors keep asking to clarify, a product feature nobody explained clearly — and each one is a one-line FAQ fix rather than a project. QaribuBot stores every thread on [Conversations](/docs/view-and-flag). Set a recurring reminder — even ten minutes a week is enough for most small sites — and treat it like checking any other channel's inbox. The value compounds: each fix you make stays fixed for every future visitor who asks the same thing. #### Re-crawl after big content changes If you launch new products or rewrite key pages, make sure the bot re-crawls so its answers stay current. QaribuBot uses the refresh schedule shown for the account's current plan, so the timing can vary by package. For a genuinely major change — a full site relaunch, a pricing overhaul — it's worth manually triggering a re-crawl rather than waiting for the next scheduled one, so the bot isn't confidently repeating outdated information in the meantime. #### What 'good training' looks like after a month By the end of the first month, a well-maintained chatbot should be answering the vast majority of routine questions correctly on the first try, with a small, shrinking list of FAQ entries covering the handful of gaps your original content didn't clearly address. That's the realistic target — not a perfect chatbot on day one, but a steadily improving one that gets better every week you spend the ten minutes reviewing it. Source: https://qaribubot.com/blog/how-to-train-ai-chatbot ### Best AI Chatbot Platforms for Customer Support Customer support has different requirements than a marketing chatbot — here's what actually matters for that use case. A chatbot built for lead capture and a chatbot built for support look similar on the surface — both are chat widgets, both use AI, both sit in the corner of your site — but they need different strengths underneath. If support is your main goal, the features that matter for a marketing-focused tool (aggressive pop-ups, lead scoring, CRM push) are secondary at best. Here's what to prioritize instead. #### Accuracy over personality A support chatbot that's charming but wrong erodes trust fast, and it erodes it in a way that's hard to win back — a visitor who gets a confidently wrong answer about a return policy doesn't just distrust the bot, they distrust your business. Prioritize platforms that ground answers in your actual content rather than generating plausible-sounding but unverified responses. This is worth testing directly before committing: ask a support-shaped question you know the answer to, and check whether the bot's answer matches your actual policy exactly, or just approximately. 'Approximately right' is a real problem in a support context in a way it might not be for casual browsing questions. #### Clear escalation paths Support questions sometimes need a human — a shipping issue that's genuinely gone wrong, an account problem, a complaint that needs empathy a script can't provide. The best platforms make it easy to hand off to email, WhatsApp, or a contact form when the bot doesn't have a confident answer, rather than leaving the visitor stuck repeating themselves or rephrasing the same question hoping for a different result. Check specifically whether the handoff carries context — does the human agent see what was already discussed, or does the visitor have to start over? A handoff that loses context is only marginally better than no handoff at all. #### Visibility into conversations You need to see what's actually being asked and answered — not just a satisfaction score — to catch problems and identify gaps in your documentation. This is doubly important in a support context, where a wrong answer has real consequences: a customer who acts on bad information about a warranty or a return window can end up genuinely worse off, not just mildly inconvenienced. Look for a dashboard that shows full transcripts, filterable or searchable if possible, so you can spot-check recent conversations without reading every single one line by line. #### Multi-site support if you run more than one If your business has several properties — a main site, a support portal, a regional site for a different market — a platform that covers multiple domains on one account, with usage centralized, is simpler to manage than separate subscriptions with separate logins, separate billing, and separate dashboards to check. #### A quick evaluation checklist Before choosing, run each finalist through the same short test: ask it three real support questions, one edge case it shouldn't know, and check the transcript afterward. This takes fifteen minutes per platform and reveals more than any spec sheet. - Does it answer real support questions correctly, from your actual policies? - Does it admit uncertainty on the edge case, rather than guessing? - Is there a visible, working path to a human? - Can you read full transcripts, not just aggregate stats? - Does it cover every domain you need under one account? Source: https://qaribubot.com/blog/best-ai-chatbot-platforms-customer-support ### How to Measure AI Chatbot Performance on Your Website Total conversation count tells you almost nothing on its own — here's what to actually track. It's easy to look at 'X conversations this month' and feel like the chatbot is working. That number alone doesn't tell you if it's helping anyone — a hundred conversations where the bot gave a wrong answer every time is a hundred failures, not a hundred successes, even though the dashboard would show the same conversation count either way. A few better metrics get you closer to the truth. #### Resolution without escalation What share of conversations end without the visitor needing to email or call you afterward? This is the clearest single signal that the bot is actually resolving things, not just chatting pleasantly without landing on a real answer. You won't always have a clean automatic way to measure this, but you can approximate it: if support email volume for a specific topic drops noticeably after the chatbot goes live and starts confidently answering that topic, that's a real, measurable win worth tracking month over month. #### Repeated questions with bad answers Read transcripts periodically and look for the same question getting a weak or wrong answer more than once. Each one is a specific, fixable gap — usually a missing FAQ entry or an outdated page that needs updating — and fixing it prevents every future visitor from hitting the same wall. This is a more actionable metric than a satisfaction score, because it tells you exactly what to fix, not just that something might be wrong somewhere. #### Usage against your quota Rising chat volume against your plan's request cap is a good problem to have — it means visitors are actually engaging with the widget rather than ignoring it. Track it so you're not caught off guard by your widget going quiet when a cap is hit partway through a busy month, which is a worse outcome than proactively upgrading a plan tier. #### Downstream signal, not just chat metrics If you can, connect chatbot activity to what happens next — did the conversation happen before a purchase, a signup, a booking? Even rough correlation is more useful than conversation count alone, since it starts to answer the question that actually matters: is this thing contributing to the business, or just generating chat logs? A simple version of this: check whether pages with the chatbot active have a meaningfully different conversion rate than they did before, or than similar pages without it. It won't be scientifically rigorous, but a clear directional signal over a month or two is genuinely useful. #### Set a monthly review habit The businesses that get the most out of a chatbot tend to have a light monthly review habit: skim usage numbers, read a handful of transcripts, note any repeated gaps, and make a small number of fixes. This doesn't need to be a formal process — a fifteen-minute check-in once a month, on the same day as your other regular business reviews, is enough to keep things improving steadily. Source: https://qaribubot.com/blog/how-to-measure-ai-chatbot-performance ### Why Should Your Website Have an AI Chatbot? Setting hype aside, here's the practical case for a chatbot on your website. You don't need a chatbot because everyone else has one, and 'because it's trendy' is a bad reason to add anything to your site. The actual reasons come down to a handful of concrete, unglamorous problems most business websites share — problems that a chatbot solves cleanly once you look past the AI marketing language around it. #### Most visitor questions are repetitive The majority of what people ask — hours, pricing, delivery, how something works — is the same handful of questions over and over, day after day, month after month. Answering them instantly, at any hour, frees your team from repeating themselves endlessly and gives visitors a faster answer than a contact form or an email ever could. If you've ever looked at your support inbox and noticed the same three or four questions making up half your volume, that's the exact pattern a chatbot is built to absorb. #### It reduces drop-off from unanswered questions A visitor with one unanswered question often just leaves rather than emailing and waiting for a response that might come tomorrow. Uncertainty is a silent conversion killer — someone who isn't sure whether you deliver to their area, or whether an item is in stock, frequently just closes the tab instead of taking the extra step to find out. A chatbot that can answer immediately keeps them on the page instead of losing them to that moment of doubt, at exactly the point where they were engaged enough to ask a question in the first place. #### It's cheap relative to the alternative Staffing live coverage for the same hours a chatbot covers costs meaningfully more, and a chatbot doesn't need breaks, training, sick days, or turnover management. For a business weighing whether to hire additional support coverage versus deploying a chatbot for the routine share of questions, the chatbot is almost always the more efficient first step. #### It surfaces what your content is missing An underrated benefit: reviewing chatbot conversations reveals questions your site doesn't clearly answer anywhere — gaps you might never have noticed without visitors actually asking. This turns the chatbot into a feedback loop for your own content, not just a support tool. Source: https://qaribubot.com/blog/why-your-website-needs-ai-chatbot ### Common Mistakes When Setting Up Website Chatbots The setup mistakes that quietly turn a good idea into an annoying widget nobody uses. Most chatbot complaints trace back to a handful of avoidable setup mistakes, not a fundamental flaw in the technology itself. The tool usually isn't the problem — the way it was configured, or never revisited after launch, is. Here are the mistakes that come up most often, in roughly the order they tend to bite. #### Letting content go stale If you change your pricing or policies and never let the bot re-crawl, it'll confidently give outdated answers — which is arguably worse than giving no answer, since a visitor has no reason to doubt a confident-sounding response. Make sure your platform re-crawls automatically on some schedule, and re-trigger it manually after major content changes rather than assuming it'll catch up on its own timeline. This mistake compounds silently: a small pricing error might go unnoticed for weeks, quietly quoting the wrong number to every visitor who asks, until someone finally flags it. #### No path to a human Trapping a frustrated visitor in a loop with no way to reach a person is the fastest way to turn a helpful tool into a source of real complaints. Always have a visible way out — email, phone, or a handoff — and don't bury it three menus deep where a frustrated visitor won't find it. This mistake is especially damaging because it happens exactly when a visitor is already annoyed — the chatbot failed to answer, and now they can't even find a way around it. That's the worst possible moment for a dead end. #### Turning it on and never checking transcripts A chatbot isn't 'set and forget.' Reviewing real conversations occasionally is how you catch bad answers before they've been repeated to fifty visitors — the businesses that get the least value from a chatbot are almost always the ones that installed it once and never looked at what it was actually saying afterward. This doesn't need to be a heavy process. Even a quick monthly skim of recent transcripts catches the majority of issues before they become a pattern. #### Overpromising in the welcome message If your greeting implies the bot can do things it can't — process a return, check an order status, override a policy — visitors will be frustrated when it can't follow through on what it seemed to promise. Keep the framing honest about what it's actually good at: answering questions from your site, not acting as a full self-service portal unless it's actually integrated to do so. #### Ignoring how it looks on mobile A widget that looks fine on a desktop browser but covers half the screen or breaks the keyboard layout on a phone is a mistake that's easy to make and easy to miss, if most of your own testing happens on a laptop. Given that most site traffic for many businesses is mobile, this is worth checking directly on an actual phone, not just assuming it translates. Source: https://qaribubot.com/blog/common-mistakes-setting-up-website-chatbots ### How to Make Your Website Chatbot Sound Natural and Helpful Small tone choices make the difference between a chatbot that feels helpful and one that feels like a wall. Accuracy matters most — a chatbot that's warm but wrong is still a bad chatbot — but tone is what determines whether visitors actually enjoy using the thing, or tolerate it at best. A few adjustments, none of them technical, go a long way toward making a chatbot feel like a natural extension of your business rather than a bolted-on tool. #### Match your brand's actual voice A formal law firm and a casual skincare brand shouldn't sound the same in chat, any more than they'd write the same email to a customer. Most platforms let you set a tone — formal, professional, casual — so replies match how you'd actually talk to a customer if you were answering yourself. This is worth doing deliberately rather than leaving on a default setting. Read a sample conversation back and ask: does this sound like us, or does it sound like a generic customer-service script that could belong to any company? #### Keep answers short by default Chat is not an essay format, and most visitors are scanning, not reading closely. A concise, direct answer with an offer to go deeper — 'want more detail on that?' — reads as more helpful than a long paragraph most people won't finish reading before losing patience and closing the widget. If a question genuinely needs a longer answer, breaking it into a short direct response followed by an offer to elaborate usually serves visitors better than a wall of text up front. #### Say 'I don't know' honestly Ironically, admitting uncertainty makes a bot sound more trustworthy, not less. A bot that clearly flags what it's unsure about, and offers a next step — a link to a relevant page, a way to reach a human — feels more natural and more credible than one that always sounds confident, even when it's guessing. Visitors are more forgiving of 'I'm not sure about that specific detail, but here's how to find out' than they are of a wrong answer stated with total confidence. #### Small details that add up The welcome message, the widget's color scheme, and how the bot signs off on a conversation all contribute to whether it feels intentional and on-brand or like a generic plugin someone installed and forgot about. None of these take long to adjust, and together they shape the overall impression far more than any single big feature would. QaribuBot puts name, greeting, color, position, tone, and persona on one Widget page — [see customization options](/docs/customize-appearance). Source: https://qaribubot.com/blog/how-to-make-chatbot-sound-natural ### AI Chatbots for E-Commerce: Product Questions and Support For online stores, the questions that go unanswered are often the reason a cart gets abandoned. E-commerce has a specific version of the chatbot problem: a shopper with one unanswered question about sizing, delivery, or returns often leaves rather than emailing and waiting. Answering those questions on the page can help people finish what they started. #### Answering the questions that stall a purchase 'Do you deliver to my area?' 'What's your return policy?' 'Is this in stock?' — these are the questions a chatbot that has read your product and policy pages can answer while someone is still on the page, rather than after they have already left to search elsewhere. The timing matters as much as the accuracy here. A shopper who has to leave your product page to find a shipping FAQ, read it, and come back has already lost momentum — and a meaningful percentage never return. Answering in place, without a page change, keeps that momentum intact. #### Reducing repetitive support load Order-status and policy questions eat a disproportionate share of support time for online stores, often more than the genuinely complex issues that actually need a human's judgment. Handling the repetitive share automatically frees your team for the cases that actually need a person — a damaged item, a genuine complaint, a bulk order with special requirements. Stores that track this often find the same five or six questions make up the majority of their support volume: shipping timelines, size guides, stock availability, return windows, and payment methods. Every one of these is a strong candidate for automatic, instant answers. #### Works alongside your existing store platform A snippet chatbot installs on Shopify, WooCommerce, or a custom store without needing a rebuild — it reads what's already on your product and policy pages rather than requiring you to duplicate that information into a separate system. Keep those pages current, recrawl when they change, and review chats when an answer is off. #### What good looks like for a store A well-set-up e-commerce chatbot should be able to answer, on its own, questions about your shipping zones and timelines, your size or fit guidance if relevant, your return and exchange policy, whether a specific product is currently in stock, and your accepted payment methods — the handful of questions that come up on nearly every product page, for nearly every visitor, at some point in their decision. - Shipping zones, timelines, and costs - Sizing, fit, or specification details - Return and exchange policy specifics - Current stock status for specific products - Accepted payment methods, including local options Source: https://qaribubot.com/blog/ai-chatbots-for-ecommerce ### What Questions Should You Program Into Your Chatbot? You don't need to write hundreds of Q&As — just the handful your site doesn't already answer clearly. If your chatbot crawls your site automatically, most content is already covered without any manual work on your part. The FAQ entries genuinely worth writing by hand are the specific gaps that crawling can't fill — and figuring out which ones those are is a smaller job than it might sound. #### Questions with no clear page Things customers ask that aren't written down anywhere on your site — an informal policy, a common clarification you usually explain by email, an exception you make more often than you'd think — are the first candidates for a manual FAQ entry. These are the cases where the crawler has nothing to find, no matter how good it is, because the answer simply doesn't exist on a page yet. A useful exercise: think about the last five questions you personally answered by email or WhatsApp this month. If they weren't clearly answered on your site, they belong on this list. #### Questions that come up in support, not on your site Check your email or WhatsApp history for the questions that repeat. If five customers asked the same thing last month, that's worth adding explicitly rather than hoping the crawl covers it well — because if it's not written anywhere, the crawl genuinely can't cover it, no matter how sophisticated the underlying AI is. This kind of audit is worth doing once at setup and then revisiting every couple of months, since the questions that repeat tend to shift as your business changes — new products bring new questions, seasonal changes bring seasonal ones. #### Anything you'd rather the bot phrase carefully Sensitive topics — refund exceptions, delays, legal disclaimers, anything where the exact wording matters — are worth writing precisely rather than trusting a general crawl to phrase them the way you want. This is less about the AI getting the substance wrong and more about control: for anything where phrasing carries legal or reputational weight, write it yourself. #### What not to bother writing manually Anything already clearly covered on an existing page doesn't need a duplicate manual entry — that's redundant effort, and it risks the manual entry and the source page drifting out of sync over time if you update one and forget the other. Trust the crawl for content that's already published and accurate; reserve manual entries for genuine gaps. Source: https://qaribubot.com/blog/what-questions-to-program-into-chatbot ### Are AI Chatbots Legal for Customer Interactions? AI chatbots are legal to use for customer interactions in the vast majority of jurisdictions — but a few practical points matter. This isn't legal advice for your specific business — regulations vary by country, industry, and how you're using the tool — but here's the general landscape most site owners should know before deploying a chatbot, and the handful of practical points that come up most often. #### Disclosure is the main requirement in most places Increasingly, regulations and platform norms expect you to make clear a visitor is talking to an AI, not a human, especially if asked directly or in certain regulated contexts. Most chatbot widgets label themselves as AI by default, often right in the greeting message — keep that labeling visible rather than removing it in the name of a more 'human' feel. This trend is only getting stronger as AI chatbots become more common and regulators pay closer attention to consumer-facing AI generally. Building disclosure in from the start, rather than retrofitting it later, is the lower-friction path. #### Data handling matters more than the AI itself The bigger legal consideration is usually what happens to the data visitors type into the chat — how it's stored, who can access it, whether it's used for anything beyond answering the immediate question, and whether your privacy policy covers it. This is true regardless of whether AI is involved at all; it's the same data-handling diligence you'd apply to any form on your site that collects visitor input. If your business operates in a jurisdiction with specific data protection requirements — GDPR-style rules, or local equivalents — those requirements apply to chat data the same way they apply to any other personal data you collect. #### It doesn't replace legally required disclosures If your industry has specific mandatory disclosures — financial terms, medical information, legal notices — those still need to appear through their proper, required channels; a chatbot answering informally in conversation isn't a substitute for a legally compliant disclosure document or page. A chatbot can point a visitor to the right formal document, but it generally shouldn't be relied on as the disclosure itself, particularly for anything regulators specifically require to be presented in a defined way. #### Liability considerations to keep in mind If a chatbot gives a visitor incorrect information that leads to a real decision — a wrong price, a wrong policy detail — think through, in general terms, how you'd want to handle that if it happened, the same way you would for a human staff member's mistake. Keeping the bot grounded in accurate, current content minimizes this risk substantially, but it's worth having a plan rather than assuming it'll never come up. #### When in doubt, ask a lawyer familiar with your jurisdiction Regulations vary by country and industry, and this article isn't a substitute for advice specific to your business, your location, and your sector — particularly if you're in a heavily regulated industry like finance, healthcare, or legal services, where the bar for automated customer interactions can be meaningfully higher. Source: https://qaribubot.com/blog/are-ai-chatbots-legal ### How to Embed an AI Chatbot Widget on Your Website Integration usually means one snippet, regardless of what your site is built on — here's what differs by platform. The good news: embedding an AI chatbot widget is usually one script tag, so the step looks almost identical no matter what your site runs on. The differences are mostly about where you paste that snippet — theme footer, custom-code box, or Google Tag Manager — not about a rebuild. For exact WordPress, Shopify, Wix, and Webflow clicks, see the [platform guide](/docs/install-on-platforms) and the [installation docs](/docs/install-the-widget). #### WordPress Paste the script into a header/footer plugin, or your theme's custom-code section if your theme supports it directly. There are no plugin conflicts to worry about since it's just a script tag, not a competing plugin ecosystem trying to hook into the same WordPress functions other plugins use. If you're using a page builder like Elementor or Divi, most of them also have a dedicated 'custom HTML' or embed block you can drop the snippet into directly on the page, as an alternative to a site-wide header/footer plugin. #### Shopify, Wix, Squarespace Each has a built-in 'custom code' or 'embed' section in the site settings specifically meant for third-party scripts like this — usually under Settings, Advanced, or Theme Code, depending on the platform. This is the same mechanism these platforms use for adding things like live-chat widgets or analytics scripts, so if you've added any third-party tool before, the process will feel familiar. #### Custom-built or framework-based sites Drop the script tag before the closing tag in your HTML template, or the equivalent root layout file if you're using React, Vue, Next.js, or a similar framework. If your site uses server-side rendering, the tag typically belongs in whatever shared layout file wraps every page, so it loads consistently across your whole site rather than just one page. #### Google Tag Manager, if you use it If your site already uses Tag Manager for other scripts, you can add the chatbot the same way — one more tag, no direct code changes to your site needed, and it can be managed by whoever already handles your other tags without touching the site's actual codebase. #### What to check after installing, regardless of platform Once the tag is in place, confirm the widget actually appears on a live page (not just a preview), check it on mobile, and send it a couple of test questions. This final check takes two minutes and catches the rare case where a script got pasted in the wrong place or blocked by a caching plugin. Source: https://qaribubot.com/blog/how-to-integrate-ai-chatbot-with-website ### Mobile-Friendly AI Chatbots: Why They Matter If most of your traffic is on a phone, your chatbot's mobile experience is effectively its only experience. For most websites, more than half of traffic is mobile, and for some businesses — anything visual, anything impulse-driven, anything discovered through social media — it's considerably more than half. A chatbot that works well on desktop but awkwardly on a phone is, practically speaking, broken for most of the visitors who'll ever encounter it, even if it looks flawless in every screenshot on the vendor's marketing page. #### What to check Does the widget resize properly on a small screen without covering the whole page or blocking the checkout button? Does the on-screen keyboard pushing up the viewport break the layout, hiding the send button or the most recent message? Test on an actual phone, not just a browser's device-simulation mode, which doesn't always replicate real mobile keyboard and viewport behavior accurately. It's worth testing on more than one device if you can — an older Android phone and a recent iPhone can render the same widget noticeably differently, especially around keyboard behavior and safe-area insets near notches or home indicators. #### Load speed matters more on mobile Mobile connections are more variable than a home or office wifi connection — a widget that adds meaningful load time can hurt your site's overall performance and search ranking, since page speed is itself a ranking factor and mobile users are quicker to abandon a slow-loading page. Look for a lightweight script, not a heavy embedded iframe that pulls in its own large bundle of assets. A well-built widget should load asynchronously, meaning it doesn't block the rest of your page from rendering while it loads — this is worth confirming directly, since a poorly implemented one can visibly delay your whole page. #### Touch-friendly, not just responsive Buttons and the send action need to be comfortably tappable with a thumb, not sized for a mouse cursor with pixel-precise clicking. This is easy to overlook if you only test on desktop during setup, where everything feels fine because a cursor is far more precise than a fingertip. Small details matter here: adequate spacing between tappable elements, a large enough input field that doesn't require zooming to read what you've typed, and a send button that's easy to hit without also tapping something else by accident. #### Why this is worth checking before, not after, launch A broken mobile experience is invisible if you only ever test on your own laptop — you can go weeks without noticing, while a majority of your actual visitors quietly struggle with it. Spending five minutes on a real phone before considering setup complete catches this class of problem entirely, and it's five minutes well spent given how much of your traffic it likely affects. Source: https://qaribubot.com/blog/mobile-friendly-ai-chatbots ### Free vs. Paid AI Chatbots: What's the Difference? Free chatbot plans are usually genuinely usable — the difference is mostly about volume, not missing features. Unlike some software categories where 'free' means crippled or watered down to the point of uselessness, chatbot free tiers are often fully functional at small scale — the real distinction is how many conversations you can handle before hitting a monthly cap, not a fundamentally worse version of the product hidden behind a paywall. #### What free usually includes A working widget, an initial site crawl, and a bounded request allowance let you evaluate the product on your own site. QaribuBot includes a one-time free trial — duration, website allowance, and request cap are shown on the pricing page and at checkout. Start the trial, connect a real page, and ask the questions your visitors actually ask. That is the fastest way to see whether the chatbot is ready for your site. #### What paid plans add Mostly higher request, website, crawl, and team capacity, plus implemented controls such as transcript export at eligible tiers. The current QaribuBot plan cards are generated from the same configuration the server enforces. This is worth confirming with any vendor you're evaluating: does the paid plan actually give you a smarter chatbot, or just a bigger allowance? If a vendor implies the free plan is deliberately worse in quality, that's a different, less trustworthy pricing model than one based purely on volume. #### When to upgrade The clear signal is hitting your monthly cap regularly — that means visitors are actually engaging, and it's worth paying to keep the widget answering rather than going quiet for the rest of the month right when it's proving useful. Waiting until you're already over the limit means visitors experience a gap in service at exactly the moment demand is highest. A good practice is checking your usage dashboard periodically, well before you're near the cap, so an upgrade is a proactive choice rather than a scramble after the widget has already stopped responding for a stretch of the month. #### What downgrading looks like, if you ever need to If your traffic drops or your needs change, most platforms let you move back down to a lower tier or the free plan without losing your setup — your crawled content, your FAQ entries, your domain connections generally stay intact. It's worth confirming this specifically, since a vendor that makes downgrading painful is using friction rather than value to keep you on a paid plan. Source: https://qaribubot.com/blog/free-vs-paid-ai-chatbots ### How to Handle Customer Complaints Through Your Website Chatbot A frustrated visitor stuck talking to a bot with no way out is a worse experience than not having a chatbot at all. Complaints are the highest-risk conversation type for a chatbot to mishandle, precisely because the visitor is already frustrated before the conversation even starts. A few deliberate design choices keep a complaint from becoming worse, and getting this right matters more than almost any other aspect of chatbot setup. #### Recognize when to step back A good setup treats complaint-shaped language — refund, broken, angry, unacceptable, this is ridiculous — as a signal to offer a human path immediately rather than attempting to fully resolve it in chat with a scripted response. The goal in these moments isn't to solve the problem in the widget; it's to move the person toward someone who can. This requires the chatbot to recognize emotional or urgent language as a distinct category from routine questions, and respond differently — with acknowledgment and a path forward, rather than the same style of factual answer it would give to 'what are your hours.' #### Never argue with a frustrated visitor The bot should acknowledge the issue plainly and move toward a resolution path, not debate the visitor's account of what happened or defend the company's position the way a poorly trained script might. This is a case where a short, honest response beats a long, defensive one — 'I'm sorry that happened, let me connect you with someone who can help directly' does more good than a paragraph explaining the policy. Visitors in this state aren't looking for a policy explanation from a bot — they're looking for confirmation that a real person will actually address the problem. #### Make escalation genuinely easy A visible email, phone number, or handoff option — not buried three menus deep, not requiring the visitor to type 'human' three times before it works — is the difference between a contained situation and an angrier one. The friction of finding a way to a person compounds an already frustrated visitor's frustration. If you offer a live-chat handoff to a human, make sure it's genuinely available during the hours you claim — an escalation path that leads to an unmonitored inbox during off-hours isn't much better than no path at all, though it's still better than a dead end. #### Follow up on what the transcript shows After a complaint gets escalated, it's worth checking the transcript afterward — not to second-guess the visitor, but to see whether the bot's initial response made things better or worse, and to catch any pattern of complaints about the same underlying issue that might need addressing at the source, not just in chat. Source: https://qaribubot.com/blog/how-to-handle-complaints-through-chatbot ### AI Chatbots for Lead Generation: Does It Really Work? Lead generation works when the chatbot removes friction at the exact moment someone's deciding — not when it interrupts. Chatbots aren't a lead-gen magic trick, and any claim that a chatbot alone will double your lead volume overnight deserves skepticism. QaribuBot answers questions from your website content and can show Call, Email, or WhatsApp buttons when those details are on your pages — it does not capture leads into a CRM or fill forms for you. What it can do is remove friction when a visitor already has a real question. #### Where it actually helps A visitor with a specific question — 'is this unit still available,' 'do you serve my area,' 'can you do this for a rush order' — who gets an instant, accurate answer is far more likely to leave contact details or take the next step than one who hits a static FAQ page, doesn't find their exact question answered, and leaves confused or unconvinced. This works because it removes a specific point of friction right when someone's motivated: they're already interested enough to ask, and an instant answer keeps that momentum going instead of letting it cool off while they search for the information elsewhere or simply give up. #### Where it doesn't A chatbot that interrupts browsing with an aggressive pop-up before a visitor has even looked around tends to hurt more than help — treat it as available, sitting quietly in the corner, not pushy and demanding attention the moment someone lands on the page. The lead-gen value comes from being helpful when asked, not from forcing an interaction. It also doesn't work as a substitute for having genuinely useful content on your site in the first place. A chatbot can't generate interest that isn't there — it can only capture and convert interest more efficiently once it exists. #### How to measure it honestly Look at whether conversations correlate with follow-up contact or bookings, not just conversation count, which tells you nothing about whether those conversations actually led anywhere. If leads aren't improving after a reasonable trial period — a month or two of real traffic — the issue is usually placement, tone, or content gaps, not the fundamental concept of using a chatbot. A useful diagnostic: read through a sample of conversations that didn't lead to a follow-up and see whether the bot actually answered well. If it did, and leads still aren't following, the bottleneck might be elsewhere in your funnel, not the chatbot itself. Source: https://qaribubot.com/blog/ai-chatbots-for-lead-generation ### How Long Does It Take to Set Up an AI Chatbot? For most sites, this is a same-day project, not a multi-week implementation. Enterprise chatbot deployments — the kind involving a sales team, a project manager, and a multi-phase rollout plan — can genuinely take weeks or months. A script-tag chatbot for a typical small business website is a different scale of project entirely. This article is only about time-to-live; the full add-a-chatbot sequence is in How to Add an AI Chatbot to Your Website. #### The realistic breakdown Signing up and registering your website: a couple of minutes. Pasting the snippet: a couple more, assuming you have access to your theme or a header/footer plugin. The first content crawl: a few minutes, depending on site size — a five-page business site finishes fast, a large catalog takes longer. Testing it with a handful of real questions: fifteen minutes if you're being thorough about it. Add that up and many small sites finish the hands-on work the same day. The chatbot can answer from what has already been read; wait for the first crawl to finish before you treat it as fully set up. #### What actually takes longer Not the setup — the refinement. Reviewing early conversations and adding a few FAQ snippets for gaps you notice is worth spending real time on over the first week or two, even though the bot is technically live and functional from day one. This isn't a blocker to launch; it's ongoing improvement that happens in parallel with the bot already doing useful work. Think of it less like a project with a hard finish line and more like publishing a page and then editing it based on feedback — except the 'feedback' here is real visitor questions showing up in your transcripts automatically. #### What could make it take longer The main things that add real time: a very large site with thousands of pages, which takes longer to fully crawl; a content security policy that needs an admin to allowlist the script; or genuinely thin site content that needs writing before the chatbot has much to work with in the first place. None of these are common for a typical small business site, but they're worth knowing about if your situation is unusual. Source: https://qaribubot.com/blog/how-long-to-set-up-ai-chatbot ### What Personal Information Should You Never Share With a Chatbot? A website chatbot is a support tool, not a secure vault — treat it accordingly. Whether you're a visitor using a chatbot or a business deploying one, it's worth being clear about what shouldn't go through this channel. Chat is convenient precisely because it's low-friction — but that same low friction can make it tempting to type things into a chat box that really belong in a secure form instead. #### Never type these into any website chatbot Full card numbers, passwords, national ID or passport numbers, and full account credentials should never be typed into a chat widget — no legitimate business needs them that way, and a chat log is not designed as secure storage for sensitive data the way a proper payment processor or authentication system is. This holds even if the chatbot seems to be asking for it directly — a legitimate business's chatbot should never request this information through chat, and if one does, that's a red flag about the business, not a reason to comply. #### What's usually fine General questions, a first name, an order number for reference, or a general location for a delivery question are typically fine — this is exactly the kind of information the chat is actually designed to help with, and sharing it doesn't create meaningful risk. The general test: would you be comfortable if this specific piece of information were visible to a support person reading a transcript later? Most everyday chatbot questions clear that bar easily; financial and identity credentials don't. #### For site owners: don't ask for more than you need If your chatbot flow asks for anything sensitive, redirect that step to a proper secure form or your existing checkout/account system instead of collecting it through chat. This isn't just good practice for visitor trust — it also reduces your own liability, since you're not storing sensitive data in a system that wasn't built to protect it. Review your chatbot's flows periodically with this lens: does any step ask for something it genuinely doesn't need to answer the question at hand? If so, that's worth removing. #### A simple rule of thumb for visitors If a human agent at the business wouldn't normally ask for it over a live chat window, don't type it into the AI version either. Businesses that need sensitive information — payment details, ID verification — almost always have a dedicated, secure flow for that, separate from casual chat. Source: https://qaribubot.com/blog/personal-info-never-share-with-chatbot ### How to Transition From Chatbot to Human Support The handoff moment is where most chatbot frustration comes from — and it's entirely fixable with good design. Every chatbot eventually hits a question it can't answer well, no matter how good the underlying content is — that's simply the nature of covering the long tail of everything a visitor might ask. What happens next, in that specific moment, determines whether the visitor feels genuinely helped or stuck with nowhere to go. #### Detect the moment early A good handoff doesn't wait for the visitor to get frustrated and ask for a human explicitly, typing 'I want to talk to a person' after several failed attempts. It recognizes low-confidence answers or repeated rephrasing of the same question as a signal to offer an alternative sooner, before frustration has a chance to build. This is one of the clearer signs of a well-designed chatbot versus a poorly designed one — the good ones proactively offer a way out when they sense they're not landing the answer, rather than requiring the visitor to ask for it. #### Make the alternative obvious and easy A visible email address, phone number, or WhatsApp link — not buried in a menu, not requiring three additional clicks — turns a dead end into a smooth transition. If you support live chat with a human during certain hours, offer it directly and clearly rather than making the visitor guess whether it's available. The easier this path is to find, the less a failed chatbot answer damages the overall experience — a visitor who quickly finds a way to a human barely registers the chatbot's miss as a problem at all. #### Don't make them repeat themselves Where possible, carry context from the chatbot conversation into the handoff, whether that's a pre-filled email subject line summarizing the question, or a note the human agent can see when they pick up the conversation. Repeating the whole question from scratch is the single most common complaint about bad handoffs, and it undoes much of the goodwill a fast chatbot response otherwise builds. Even a simple version of this — an email template that auto-fills with 'Regarding: [the visitor's last message]' — meaningfully reduces the friction of starting over. #### Review handoffs specifically, not just chatbot answers When you're reviewing transcripts, pay particular attention to the conversations that ended in a handoff — these are your best signal for where the chatbot's knowledge genuinely runs out, and often point to specific content gaps worth closing so fewer future visitors need the handoff at all. Source: https://qaribubot.com/blog/chatbot-to-human-handoff ### Best Practices for Website Chatbot Customer Service A short, practical checklist rather than a theoretical framework. Most of what separates a good chatbot deployment from a bad one comes down to a handful of concrete, repeatable habits, not a fundamentally different or more expensive tool. The businesses that get real ongoing value from a chatbot tend to do the same small set of things consistently — here's what those things are. #### Keep source content current Re-crawl after major changes to pricing, policy, or products so the bot never confidently states something that's no longer true. This is the single highest-leverage habit, because a stale chatbot doesn't just fail to help — it actively misleads, and it does so with the same confident tone it uses for correct answers, so visitors have no way to tell the difference. #### Review transcripts on a schedule Even fifteen minutes a week, skimming recent conversations, catches problems early — before they've repeated across dozens of visitors and become an ongoing source of quiet frustration nobody's tracking. Put it on a calendar the same way you would any other recurring business task, rather than relying on remembering to check. #### Always offer a way to a human No exceptions here — this is the single most important safety net for a chatbot deployment, and skipping it turns every chatbot limitation into a dead end instead of a manageable gap. #### Set a tone that matches your brand A mismatched tone — too stiff for a casual brand, too casual for a formal one — undermines trust even when the answers themselves are perfectly accurate. Tone is one of the fastest things to adjust and one of the easiest to leave on a generic default, which is exactly why it's worth deliberately setting rather than skipping. #### Test on mobile, not just desktop Given how much traffic for most businesses is mobile, confirming the widget genuinely works well on a phone — not just a browser's simulated mobile view — is worth doing directly rather than assuming. #### Treat it as a living tool, not a one-time install The businesses that get the most value tend to think of their chatbot the way they think of their website itself: something that improves over time with attention, not something you set up once and never revisit. None of the individual habits above take much time — the value comes from doing them consistently. Source: https://qaribubot.com/blog/best-practices-chatbot-customer-service ### AI Chatbots and Customer Privacy: What You Need to Know For site owners: the privacy questions worth answering before you turn a chatbot on. Adding a chatbot means a new place where visitor data flows through your systems — every question a visitor types is, technically, data you're now collecting and storing somewhere. A few questions are worth answering upfront rather than after the fact, when a visitor or a regulator asks and you don't have a ready answer. #### Know where conversation data is stored Ask your chatbot vendor plainly: where does chat data live, who can access it, and for how long is it retained? You should be able to answer this confidently if a visitor or regulator asks, rather than having to go find out after the fact. A vendor that can't answer this clearly themselves is a warning sign about their own data practices. This is a reasonable question to ask before signing up for any chatbot platform, not just after you're already using one — it belongs in your evaluation checklist alongside pricing and setup time. #### Update your privacy policy If you're collecting chat conversations, your privacy policy should mention it — most site owners overlook this when adding a new tool, treating it as a technical install rather than a change to what data they're collecting. A short addition covering what's collected through chat and roughly how it's used closes this gap. This is a quick fix once you're aware of it, and it's the kind of detail that matters more in the aggregate — across every tool your site uses — than any single tool in isolation, but it's worth getting right for each one. #### Don't design flows that collect more than needed If your chatbot doesn't need a visitor's phone number to answer their question, don't build a flow that asks for it anyway, even if it seems like a convenient way to capture a lead. Collecting data you don't need adds risk without adding proportional value, and it's the kind of practice that erodes visitor trust if noticed. A useful discipline: for every piece of information your chatbot flow asks for, be able to state exactly why it's needed to answer the visitor's question or serve them better. If you can't, reconsider asking for it. #### Have a plan for data requests In jurisdictions with data protection regulations, visitors may have a right to request what data you hold about them, including chat history, or to request deletion. Know in advance how you'd fulfill such a request through your chatbot vendor's dashboard or support channel, rather than figuring it out for the first time when someone actually asks. Source: https://qaribubot.com/blog/ai-chatbots-and-customer-privacy ### How to Update Your Chatbot's Responses and Information An accurate chatbot on day one can be a wrong chatbot by month three if nobody maintains it. Most of the maintenance work is passive if your platform re-crawls automatically — but a few situations still need a manual nudge, and knowing which is which saves you from either over-managing a tool that mostly takes care of itself, or under-managing one that quietly drifts out of date. #### Automatic re-crawling handles most of it If your platform re-crawls your site on a schedule, routine content changes can be picked up without any action from you. QaribuBot uses the refresh interval configured for your current plan; the live pricing page shows the applicable schedule. See [How we learn your pages](/docs/how-crawling-works) for credits, automatic recrawls, and manual runs. #### Trigger a re-crawl after major changes For a big update — new pricing, a rebranded site, a major policy shift, a full redesign — don't wait for the next scheduled crawl if your platform lets you trigger one manually. The gap between a major change going live and the bot catching up is exactly when it's most likely to confidently state something that's no longer true, which is worse than simply being slow to update. This is a small habit worth building into your launch checklist for any significant site update: alongside checking your analytics tracking still works and your forms still submit, add 'trigger a chatbot re-crawl' to the list. #### FAQ snippets need manual updates Anything you added as a manual FAQ entry won't update itself, since it's not tied to a live page the crawler re-reads — if a policy in a snippet changes, go back and edit that entry directly, or it'll keep repeating the old version indefinitely. That lives on [Knowledge](/docs/add-knowledge-manually) in QaribuBot. It's worth keeping a mental (or literal) list of what you've added manually, precisely because these entries are the ones that won't self-correct the way crawled content does. #### A simple quarterly check Beyond the routine and post-launch updates, it's worth doing a broader review every few months: read through your manual FAQ entries and confirm they're all still accurate, and spot-check a few common questions to make sure the bot's answers still match reality. This catches the slow drift that individual updates sometimes miss. Source: https://qaribubot.com/blog/how-to-update-chatbot-responses ### Multi-Language AI Chatbots for Global Websites For sites with visitors from more than one language background, chatbot language handling is worth checking before you launch. If your visitors don't all speak the same language, it's worth understanding how your chatbot actually handles that rather than assuming it works seamlessly by default — multilingual support is one of those features that sounds simple in a product description but varies a lot in practice. #### How it typically works Modern AI chatbots generally reply in whichever language a visitor writes in, as long as the underlying model supports it — you usually don't need to configure separate language versions manually or maintain parallel content for each language your visitors use. This is a meaningful improvement over older chatbot technology, which typically required building entirely separate conversation flows per language, effectively multiplying the setup work by however many languages you wanted to support. #### The catch: source content still matters If your site content is only published in one language, the bot's ability to answer accurately in other languages depends on the model's general language ability, not your specific business content, since it can only ground answers in what it's actually read. So the safest approach for a genuinely multilingual audience is publishing key pages — pricing, policies, core product info — in every language your audience actually uses. This matters most for anything specific to your business that a general-purpose AI wouldn't know on its own, like your particular return policy or delivery zones — general knowledge questions are far less affected by this limitation than business-specific ones. #### Test with real questions in each language Don't assume — actually type in a few questions in each language your visitors use and check the answers, since quality can vary by language even within the same underlying tool. A chatbot might handle one language flawlessly and another with noticeably more errors, and the only way to know is to test directly rather than trust a generic 'multilingual support' claim on a features page. #### A practical starting point If you're not sure where to start, look at what languages your actual visitor traffic uses — most analytics tools show a language or country breakdown — and prioritize testing and, if needed, publishing content in your top two or three languages by traffic share before worrying about less common ones. Source: https://qaribubot.com/blog/multi-language-ai-chatbots ### Real-World Examples of Successful Website Chatbots Concrete patterns from businesses that use chatbots well, across a few different industries. Rather than abstract benefits, it's more useful to look at what a well-used chatbot actually looks like across a few common site types — the specific questions it's answering, and why those particular businesses see real value from it rather than treating it as a checkbox feature. #### Online stores When a shop keeps product and shipping pages current, the bot can field 'is this in stock' and 'do you ship to my area' from that material instead of leaving those questions only in a support inbox. Catalogs and policies change often, so recrawl when they do. #### Schools and educational sites Admissions and fee questions repeat every term, often from parents contacting the same office by phone dozens of times over a short admissions window. A chatbot that has read the admissions and fees pages can take those repeating questions, so staff can focus on applications that need a person's judgment. #### Service businesses and agencies A chatbot that has read the services and pricing pages can answer 'do you do X' and 'how much does Y cost' from those pages, including when nobody is at the desk to reply to a contact form. #### Real estate and property listings On listing and financing pages, a chatbot can answer after-hours questions about availability or service charges from what you published, so a late-evening visitor is not waiting until the next business day for a basic FAQ. #### The common thread In every case, the bot is only as good as the content it has read — success comes from businesses that keep their site content accurate and occasionally review real conversations, not from the technology alone doing all the work unattended. The businesses that treat it as a living tool rather than a one-time install consistently get more out of it. Source: https://qaribubot.com/blog/real-world-examples-successful-chatbots ### How to Make Your Chatbot Personality Match Your Brand A chatbot that sounds nothing like the rest of your brand feels like a bolted-on tool, not part of the experience. Tone consistency is an easy win that a lot of chatbot deployments skip entirely by leaving everything on a generic default setting — which is a missed opportunity, since it's one of the lowest-effort ways to make a chatbot feel like a genuine part of your business rather than a third-party plugin bolted onto your site. #### Start with your existing writing Look at how your website, emails, or social captions already talk to customers — formal, warm, playful, no-nonsense — and set your chatbot's tone to match rather than leaving it generic. If your brand voice already exists in your marketing copy, you're not starting from scratch; you're just extending a voice you've already defined into a new channel. A quick way to check alignment: read a sample chatbot conversation back to back with a page of your website copy. If they don't sound like they came from the same business, that's worth adjusting. #### Small details carry a lot of weight The welcome message, the color of the widget, and how it signs off all contribute to whether it feels like 'you' or like a generic plugin bolted onto your site. None of these individually seem important, but together they set the first impression a visitor forms about whether this chatbot genuinely belongs to your business. Set those in QaribuBot from [Widget customization](/docs/customize-appearance). The widget color especially is an easy, purely visual win — matching it to your brand palette rather than leaving a default takes seconds and immediately makes the widget feel intentional rather than default. #### Consistency over cleverness A chatbot personality that's fun but inconsistent with the rest of your brand can feel jarring — a playful, joke-cracking chatbot on an otherwise formal, serious site creates a mismatch that undermines trust rather than building rapport. Consistent and slightly plain beats inconsistent and quirky, especially for anything customer-facing where trust matters. #### Revisit it as your brand evolves If you ever rebrand, update your visual identity, or shift how you talk to customers elsewhere, remember to update the chatbot's settings too — it's easy to update every other customer touchpoint and forget the chat widget, leaving it as a small but noticeable holdover from an old brand voice. Source: https://qaribubot.com/blog/chatbot-personality-match-brand ### AI Chatbots for Appointment Scheduling: Complete Guide Chatbots are excellent at answering scheduling questions — booking itself usually still needs a dedicated tool. There's a useful distinction here that's easy to blur: a chatbot answering questions about booking, and a chatbot that actually books the appointment for you, are two different levels of capability. QaribuBot does not book appointments. It can answer hours and booking-page questions from your published content, and it can point visitors to a booking link when that link is on your site. #### What a content-trained chatbot handles well 'What are your hours?' 'How do I book?' 'What's your cancellation policy?' 'Do you have availability this week?' — a chatbot that has read your booking page can answer these from that page, which removes friction before someone reaches your actual booking system. This is useful on its own, even without a booking integration. For many service businesses, the questions around scheduling are as much a barrier to booking as the scheduling mechanics themselves — a visitor unsure whether you're open on weekends, or what your cancellation policy actually is, might not even get as far as your booking calendar without that reassurance first. #### Actual booking usually needs integration To let visitors book directly inside the chat — pick a time slot and confirm an appointment without leaving the widget — your chatbot needs to connect to a scheduling tool like a calendar system. Not every chatbot platform supports this out of the box, so check specifically if that's your goal rather than assuming it's included by default. This is a meaningfully more complex integration than the crawl-and-answer functionality most chatbots offer as standard, since it requires two-way communication with your booking system, not just reading content. #### The practical middle ground Most businesses get real value just from the bot answering scheduling questions and linking directly to the booking page at the right moment in the conversation — full in-chat booking is a nice-to-have, not a requirement to get started. A chatbot that says 'Yes, we have availability most weekdays — here's the link to book directly' accomplishes most of the practical benefit without needing a deep calendar integration. #### Making the handoff to your booking page smooth If you're relying on the link-out approach rather than in-chat booking, make sure your booking page itself is easy to use once someone arrives there from the chat — a clunky booking flow at the destination undoes some of the momentum the chatbot built by answering quickly in the first place. Source: https://qaribubot.com/blog/ai-chatbots-for-appointment-scheduling ### Why Customers Prefer Talking to Chatbots (And When They Don't) The preference isn't about chatbots versus humans in the abstract — it's about matching the tool to the question. Whether people like chatbots depends on the question, not a blanket preference for or against AI. Framing it as chatbots-versus-humans in the abstract misses the more useful question underneath. #### When people prefer the bot For quick, factual questions, many people would rather get an answer now than wait in a queue or until the next business day — for example opening hours or whether you ship to their area. There's also a lower-stakes comfort factor: some visitors prefer asking a bot a basic question over calling and potentially feeling like they're taking up a real person's time for something simple. #### When people want a human Anything emotionally charged, ambiguous, or high-stakes — a real complaint, a big purchase decision, something that needs judgment or an exception a policy page doesn't address — people generally want a person, and being stuck with only a bot in that moment causes real, disproportionate frustration compared to a routine question going unanswered. This is worth taking seriously as a design principle, not just an observation: the cost of a bad chatbot interaction is much higher for these conversation types than for routine ones, which is exactly why a clear escalation path matters most here. #### The takeaway for site owners Design for both: let the bot handle the fast factual layer where speed is genuinely preferred, and make the human path obvious for anything that needs it. Neither replaces the other well, and trying to force either extreme — an all-bot support model, or refusing to use a chatbot at all because 'customers prefer humans' — misses the more accurate, more useful middle ground. The businesses that get this right tend to think of it less as 'chatbot versus human' and more as routing: fast, factual, low-stakes questions to the bot; everything else to a person, with a smooth path between the two. Source: https://qaribubot.com/blog/why-customers-prefer-chatbots ### How to Monitor and Improve Your Chatbot's Conversations A ten-minute weekly habit that keeps a chatbot improving instead of quietly degrading. The single highest-leverage thing you can do after launch is a lightweight, consistent review routine — not a big analytics project with dashboards and KPIs, just a habit of actually reading what visitors are asking and what the bot is telling them. #### Set a recurring reminder Ten or fifteen minutes a week, reading a sample of recent conversations, is enough to catch most issues early. This doesn't need to be exhaustive — you're not trying to read every single conversation, just enough of a sample to notice patterns before they become widespread. Attaching this to an existing weekly habit — a Monday morning check-in, a Friday wrap-up — makes it far more likely to actually happen than relying on remembering to do it whenever you think of it. #### Look for patterns, not one-off oddities One weird answer isn't worth chasing — every AI system occasionally produces an unusual response to an unusual phrasing, and treating every single anomaly as urgent is a good way to burn out on the review habit quickly. The same question getting a weak answer three times in a week is worth a fix — usually a new FAQ entry or a content update. This distinction between signal and noise is what makes the habit sustainable long-term: you're watching for repeats, not perfection on every single exchange. #### Close the loop After you spot a gap and fix it, check back in a week or two to confirm the fix actually worked — the point of the review isn't just noticing problems, it's confirming they're resolved. It's easy to make a fix and never verify it actually addressed the underlying issue. Keeping a simple running note of what you've fixed and when also helps you notice if the same underlying issue keeps resurfacing in a different form, which sometimes points to a deeper content gap than the individual fixes alone address. #### Involve your team if you have one If more than one person handles customer questions at your business, it's worth sharing interesting or concerning transcripts with them periodically — they often notice things you'd miss, especially around whether the bot's tone and accuracy match what they'd say themselves in the same situation. Source: https://qaribubot.com/blog/how-to-monitor-improve-chatbot-conversations ### Chatbot Downtime: What to Do When Your AI Tool Goes Down Most 'chatbot is down' situations are simpler to diagnose than they feel in the moment. Before assuming the worst — a full outage, a broken integration, a serious problem — a short checklist usually finds the actual cause quickly, and in most cases the fix takes minutes once you know where to look. #### Check if it's actually the widget or your quota The most common cause isn't an outage at all — it's hitting your plan's monthly request cap. Check your usage dashboard first; if you're at the limit, the widget going quiet is expected behavior, not a bug, and the fix is either upgrading your plan or waiting for the next billing period. This is worth checking first specifically because it's by far the most common cause, and it's easy to rule out quickly rather than jumping to more complicated explanations. #### Check if the snippet is still present A site redesign, theme change, or plugin update can accidentally remove a custom script without anyone noticing immediately — especially if someone else on your team made the change and didn't know the chatbot tag needed to be preserved. Confirm the tag is still in your site's code, viewable through your browser's page source, before assuming the provider itself is down. This is a particularly common cause after a site migration or a theme switch, since custom code snippets are exactly the kind of thing that gets lost in a platform change unless someone specifically checks for it. #### Check your provider's status page A real outage on the vendor's side should be visible on their status page or through their support channel — if there's nothing reported there, the issue is more likely local to your site or account rather than a widespread problem affecting every customer. If the widget itself never appears, start with [Widget not showing up on my site](/docs/widget-not-showing). It's worth knowing where to find this before you need it, rather than searching for it for the first time in the middle of an actual issue. #### Have a fallback for extended downtime Keep a visible contact email or phone number on your site independent of the chatbot, so visitors always have a way to reach you even if the widget is temporarily unavailable for any reason. This is good practice regardless of how reliable your chatbot provider is — no single tool should be the only way for a visitor to reach you. Source: https://qaribubot.com/blog/chatbot-downtime-what-to-do ### How AI Chatbots Can Reduce Your Customer Service Costs Any saving comes from fewer repetitive questions landing on staff — only if the bot answers those questions well. A chatbot can take some of the repetitive questions that would otherwise take staff time, including outside office hours. Whether that is worth the monthly plan depends on your volume and how accurate the answers are. Do not take a generic 'AI saves money' claim on faith. #### Where the savings actually come from If a share of your support volume is repetitive factual questions, a chatbot answering those from your pages can mean less staff time on them, without hiring for after-hours coverage. That only holds when the answers are right — a wrong answer can create more work, not less. This is a direct, calculable saving, not a vague productivity claim: every question the chatbot correctly answers on its own is a question that didn't consume a staff member's time, email response, or phone call. #### A rough way to estimate it Estimate how many support interactions per month are simple, repeatable questions — check your email or WhatsApp history for a rough count — multiply by the average time a staff member spends on one (often a few minutes per email, once you count reading, writing, and any back-and-forth), and compare that freed-up time against the chatbot's monthly cost. Compare that time against the chatbot's monthly cost. Whether it pays for itself depends on your volume and how often the bot is actually right — we do not claim a typical payback period. #### The compounding effect over time Unlike hiring, extra routine questions still count against your plan's chat allowance. As traffic grows you may need a higher plan — the cost does not stay flat forever. #### How it works alongside your team The chatbot takes the routine volume — hours, pricing, policies, product details — so your team can spend time on the conversations that need a person. That is usually the highest-leverage split: instant coverage on published facts, plus humans for the high-stakes cases. Source: https://qaribubot.com/blog/how-ai-chatbots-reduce-customer-service-costs ### Website Chatbots: Industry-Specific Use Cases The same core technology, applied differently depending on what your visitors actually need to know. A chatbot's value looks different depending on your industry, because the repetitive questions worth automating are different in each one — the underlying technology is the same, but what it's actually asked to do, and what makes it useful, shifts a lot from sector to sector. #### E-commerce Product availability, shipping, and returns — questions that directly affect whether a cart gets completed. For online stores, the chatbot's value is most visible in the moments right before a purchase decision, when an unanswered question is most likely to cause someone to abandon their cart. #### Real estate Listing availability, financing questions, and neighbourhood details — freeing agents from repeating the same answers while they're out on viewings and unavailable to respond to inbound questions in real time. This is a sector where the chatbot's after-hours availability is particularly valuable, since property browsing often happens in the evening. #### Healthcare and clinics Hours, services offered, and insurance accepted — general questions, clearly kept separate from anything resembling medical advice, which a chatbot should never attempt to provide. The value here is specifically in reducing the volume of routine administrative calls a front desk fields, freeing staff for in-person patient needs. #### Education Admissions timelines and fee structures — the same questions parents ask every term, answered instantly instead of tying up the front office during what's often the busiest period of the academic calendar. Schools that deploy this well see a particularly noticeable drop in repetitive phone traffic during admissions windows. #### Government and public services Procedural questions — what documents are needed, where a specific office is located, what a process requires — are exactly the kind of high-volume, repeatable question a chatbot can absorb, reducing both phone queue times and in-person walk-in volume for routine inquiries. #### Non-profits and NGOs Program eligibility and how-to-donate questions repeat constantly for lean teams that don't have dedicated support staff to spare. A chatbot here often has an outsized impact relative to its cost, precisely because these organizations typically have the least staff capacity to absorb repetitive inquiries manually. #### The pattern across all of them In every sector, the highest-value use case is the same underlying pattern: identify the questions that repeat most often and matter least in terms of requiring individual judgment, and let the chatbot absorb those, freeing people for the conversations that genuinely need a human's expertise or authority. Source: https://qaribubot.com/blog/website-chatbots-industry-specific-use-cases ### How to Get Customer Feedback on Your Website Chatbot Direct feedback is a useful complement to reading transcripts yourself — here's how to collect it without annoying visitors. Reading transcripts tells you a lot about how your chatbot is performing, but asking visitors directly catches things you might read differently than they actually experienced — your interpretation of a conversation as 'fine' and a visitor's experience of it as 'fine' aren't always the same thing. #### A simple end-of-conversation prompt A lightweight 'was this helpful?' after a conversation wraps up gets you a usable signal without adding friction — keep it to one tap or click, not a full survey with multiple questions that most visitors will simply ignore or close out of. Even a low response rate on a simple thumbs-up/thumbs-down prompt is useful directionally over time — you're not looking for statistical rigor, just a rough signal you can track month over month alongside your transcript reviews. #### Watch for feedback given in the chat itself Visitors often say things like 'that's not what I meant' or 'this isn't helping' directly in the conversation — these are free-form feedback if you're reading transcripts, even without a formal rating system in place. This kind of unprompted, in-context feedback is often more specific and useful than a generic satisfaction score, because it tells you exactly what went wrong. Make a habit of specifically flagging these moments when you're doing your regular transcript review, since they're a concentrated source of the most actionable feedback available to you. #### Don't over-ask A rating prompt on every single conversation gets ignored fast, the same way any survey fatigue sets in when asked too frequently. Reserve it for conversations that ran a few exchanges, where the visitor is more invested in the outcome and more likely to give a real, considered answer rather than dismissing the prompt reflexively. #### Combine both sources for the full picture Neither source alone tells the whole story — explicit ratings miss the nuance of what actually went wrong, while transcript review alone misses the visitor's own sense of whether they were satisfied. Together, they give a more complete picture than either would on its own, and neither requires much additional setup once you have transcript access. Source: https://qaribubot.com/blog/how-to-get-feedback-on-chatbot ### AI Chatbot Integrations: Connect With Your Business Tools Integrations extend what a chatbot can do beyond answering questions — here's what to realistically expect. A chatbot that just answers questions from your site content is useful entirely on its own. QaribuBot does not include CRM, calendar, or booking-tool integrations. You can export conversation transcripts from eligible plans. Treat extra tool connections as a separate project, not a prerequisite for going live. #### Common integration points Sending captured leads to a CRM, connecting to a booking calendar, or exporting conversation data for analysis in a separate tool are the most common extensions businesses look for once the basics are working well and they've identified a specific, recurring need for a connection to another system. The pattern worth noting: integrations tend to be additive on top of a working core product, not a replacement for it — a chatbot with zero integrations still answers questions correctly; integrations mainly add convenience around what happens with the data afterward. #### Start simple, add integrations when there's a real need It's tempting to want everything connected on day one, especially if you're comparing feature lists across vendors and integrations look impressive on paper. But most businesses get the bulk of the value from the chatbot answering questions well first, and add integrations once a specific, recurring, well-defined need shows up — not speculatively, in case it might be useful someday. Starting simple also means you're not troubleshooting a complex integration setup on day one, when you're still learning how the core product behaves and what your own content gaps look like. #### Check what's actually supported before committing If a specific integration matters to you — a particular CRM, a specific booking tool, a specific analytics platform — confirm it's actually supported rather than assuming, since integration depth varies a lot between platforms and a vendor's general 'integrations available' claim can mean very different things in practice. Ask specifically about the integration you care about, rather than trusting a generic integrations page — some 'integrations' are deep, two-way connections, and others are little more than a webhook you'd need to build additional tooling around yourself. #### Exporting data even without a formal integration Even without a dedicated integration, most platforms let you export conversation transcripts, which you can manually review or feed into whatever analysis tool you already use. This is a reasonable middle ground if you want visibility into chatbot data without committing to a formal ongoing integration right away. Source: https://qaribubot.com/blog/ai-chatbot-integrations ### Starting Your First Chatbot Project: What You Need You need less than you'd think — here's the actual short list. If you've read this far into a chatbot blog, you've probably overthought the prerequisites already. This is a before-you-start checklist — not another how-to. For the install sequence, see How to Add an AI Chatbot to Your Website. #### A website with real content The main input is your existing pages — products, services, FAQs, policies. If your site is thin on content, the chatbot will only be as good as what it can read, so a quick content review is worth doing first: are your pricing, hours, and policies actually written down clearly anywhere, or do they only exist in your head or in past email replies? This audit alone is often useful independent of the chatbot project — it frequently surfaces gaps in your site content that were worth fixing regardless. #### Permission to add the snippet You'll need whatever access lets you paste a snippet into your site — a theme editor, a header/footer plugin, or your CMS's custom-code settings. If that's not you personally, it's a five-minute favor from whoever does have that access, not a development task requiring a project brief. #### A short list of your top visitor questions Even a rough mental list of the five questions you get asked most is enough to know what to test once the bot is live, and where to add FAQ entries if the automatic crawl misses anything. You probably already know this list without writing it down — it's whatever you find yourself explaining to customers over and over. #### A little time in the first week for review Not a large time commitment, but plan to spend fifteen or twenty minutes sometime in the first week reading through early conversations. This is where most of the genuine 'setup' value happens — not in the initial installation, which takes minutes, but in the light-touch refinement afterward. #### That's genuinely it No developer, no multi-week implementation project, and no dedicated project manager are required to get started — you can use the one-time trial shown at checkout and decide from real usage whether a paid package fits. Source: https://qaribubot.com/blog/starting-your-first-chatbot-project ## Legal - [Terms of Service](https://qaribubot.com/terms): Billing, acceptable use, and service terms. - [Privacy Policy](https://qaribubot.com/privacy): How QaribuBot handles visitor and merchant data. - [Cookie Policy](https://qaribubot.com/cookies): Essential and optional analytics cookies, and how to change your choice. - [Refund Policy](https://qaribubot.com/refund-policy): How refunds work for card and mobile money payments.