Skip to content
Back to blog

Page Targeting Chat Widget: Setup, Rules and Examples

A page targeting chat widget shows the right greeting, prompts and answers for each URL. Learn the rules, openers and checks that make it work on a small site.

A page targeting chat widget changes what it does based on the page a visitor is on. On a pricing page it opens with plan questions. On a product page it talks about that product. On a help page it points to support. You set rules by URL, path or page type, and a default behavior covers everything else. Done well, it feels like a helpful shop assistant. Done badly, it feels like the same pop-up shouting at every page.

This guide covers how page targeting works, which rules are worth setting, what to write for each page type, and how to test it before visitors see it.

Konvuno
Give Product Pages More Relevant Answers
Konvuno answers visitor questions from your website content and keeps product details current by syncing WooCommerce and supported product feeds.
Explore Konvuno

TL;DR

  • Page targeting means matching a chat experience to the current URL, path or page category, with a default for unmatched pages.
  • Start with four page groups: homepage, product or service pages, pricing or checkout, and help pages. Add more only when your analytics show a need.
  • Tell visitors what the assistant can see. Users often do not assume it knows which page they are on.
  • On product pages, answers must come from current catalog data, not static text.
  • Test the launcher and open panel on desktop and mobile, because other widgets fight for the same corner.

Table of Contents

What page targeting actually does

The mechanism is simple. The visitor loads a page, the widget checks your URL rules, and a matching assistant or flow handles the conversation. If nothing matches, a default assistant takes over. In tools that support it, rules can target paths such as /pricing or /product, and you can run several assistants with a fallback.

There are three things you can vary per page:

  1. Visibility. Show or hide the launcher. Hiding it on a checkout step or a booking form is a legitimate choice.
  2. Opening message and starter prompts. What the widget says first, and which clickable questions it offers.
  3. Knowledge scope. Which content the assistant leans on: the product in view, the shipping policy, the service description.

The third one matters most and is the one people skip. A greeting change is cosmetic. An assistant that answers from the right source for the page is useful.

If you are on WordPress, some chat plugins already let you display a widget on all pages or only on specific products, landing pages, or cart and checkout. That covers visibility. Prompts and knowledge scope are a separate job.

Why match the chat to the page

Visitors arrive with different intent depending on where they land. Nielsen Norman Group researchers Georgia Kenderova, Maria Rosala and Tanner Kohler studied real users with AI chatbots in 2026 and found that contextual relevance strongly affects whether people rely on the bot. Their guidance: product-page prompts should address the product being viewed, while homepage prompts can support broader exploration.

Different website pages mapped to relevant chat prompts

The same research warns of a quiet failure. If the interface does not say the bot can see the current page, users may copy and paste a product name into the chat. That is friction you created by being silent about what the assistant knows.

For a small business, the practical payoff is fewer dead-end conversations. A visitor on a delivery-policy page who gets “How can I help you today?” has to restate everything. A visitor who gets “Questions about delivery? Ask about costs, timing or returns” is already in the right place.

Page groups worth targeting first

Do not build forty rules. Most small sites need four to six. Here is a starting map, with the intent behind each group.

Page group Visitor intent Opening message idea Starter prompts
Homepage Exploring, not sure yet Short, open: what you do and how the assistant can help “What do you offer?”, “Where do I start?”, “Talk to a person”
Product or service page Evaluating this specific item Names the page context: “Questions about this product?” Size or fit, compatibility, availability, shipping
Pricing or plans Ready to compare and buy Offers help choosing “Which plan fits me?”, “What is included?”, “Can I try first?”
Cart and checkout Close to paying, easily derailed Often best hidden, or a tiny help prompt Delivery cost, payment, returns
Help, FAQ, shipping Solving a problem Support guidance “Track my order”, “Start a return”, “Contact support”
Contact page Wants a human Offer the fastest route to one “Leave a message”, “Request a callback”

The NN/G guidance lines up with this: broad exploratory prompts on a homepage, product-specific questions on product pages, support guidance on help pages, and buying or pricing assistance on commercial pages. They also report that HubSpot found 15% of its website chat questions were technical-support questions, which is a reminder that visitors use chat for more than sales, wherever you put it.

Writing the opening message for each page

The best page-specific greeting does two jobs in one or two sentences: it shows the assistant knows where the visitor is, and it narrows the field of useful questions.

Weak: “Hi! How can I help?”

Better on a product page: “Looking at the Alder Rain Jacket? Ask about sizing, materials, stock or delivery.”

Better on a pricing page: “Not sure which plan fits? Tell me about your site and I’ll point you to one.”

Better on a returns page: “Need to return something? I can walk you through it or connect you to our team.”

A few rules we apply:

  • Name the page context. Mention the product, the topic or the section. That is how you make awareness explicit.
  • Offer two to four clickable starters, not ten. Academic research on chatbot interaction design indicates predefined buttons can improve pragmatic and hedonic experience compared with free-text-only interaction. Keep free typing open too.
  • Always show a way to a person. A button or line that says how to reach a human should survive every page rule.
  • Do not promise what the assistant cannot answer. If it cannot see order status, do not greet with “Ask me about your order.”

For more on conversation starters and making page awareness obvious, this walkthrough of chat and visitor-identification tooling from a small-city marketing agency is a useful outside reference. For the wording of the instructions behind the assistant, see our piece on three research-backed checks for chatbot prompts.

Product pages: ground answers in your catalog

On an online shop, page targeting only works if the answers are right. A greeting that names the product followed by a wrong price is worse than no greeting.

That means the assistant should read from current catalog data, not from a block of static text someone typed last spring. Google’s merchant documentation shows what structured product data looks like: price, availability and condition are core product fields, and structured data markup on your pages lets crawlers understand the same facts. Google also offers optional conversational attributes, such as question-and-answer pairs, document links, related products and variant options, to help AI systems understand product nuances.

The operational lesson is that catalog quality is customer-facing quality. Whatever is wrong in the feed will eventually be said to a shopper. If you want a broader view of which product details to expose on product pages, this guide to optimizing product pages for sales and SEO covers the data points shoppers look for.

If you run WooCommerce or publish a Google Merchant or Facebook feed, two of our guides go deeper: getting an accurate product feed chatbot live in one day and a seven-step fix for feed errors such as bad exports and variant IDs.

Where the widget sits on the page

Page targeting also covers placement. Cookie banners, accessibility toggles, booking buttons, checkout bars and back-to-top arrows often compete for the same bottom corner. NN/G’s guidance is to test the launcher and the open panel at both desktop and mobile widths, since an overlapped control is a lost conversation.

A quick check on each targeted page group:

  1. Load the page on a phone-sized screen and a laptop.
  2. Trigger the cookie banner and confirm it does not cover the launcher.
  3. Open the panel and confirm it does not hide the add-to-cart or checkout button.
  4. Tab through the page with a keyboard and confirm the launcher is reachable.

We cover placement in more detail in chatbot button placement: rules and five metrics to test.

A perspective from the people building one

We build Konvuno for owners who have a website and no support team, so our view is shaped by what they can realistically maintain.

Page targeting sounds like a developer feature, but the hard part is not the rule. It is the content behind the rule. We see the same pattern: an owner sets up elaborate per-page greetings, then the assistant answers from stale or thin information and trust drops anyway. So we would reverse the order of work. First make the knowledge good: FAQs organized by topic, site content the assistant can read, and a product catalog that syncs on its own. Then add page-specific greetings where your analytics show visitors asking different things in different places.

A perspective from the people building one — overview diagram

That is also why we built the dashboard around what visitors actually ask. Conversations and question analytics tell you which pages need their own opening message. You do not need to guess.

We also think restraint wins. A small business with six well-written page groups beats one with sixty half-maintained ones. Every rule is something you must remember to update when a page, a price or a policy changes.

— Konstantin

A rollout plan you can finish in a week

  1. Day 1: pull the data. List your top ten pages by traffic and your top ten questions customers email or call about. Match each question to the page where it arises.
  2. Day 2: group pages. Sort into the groups from the table above. Anything that does not fit goes to the default.
  3. Day 3: write greetings and starters. One greeting and three starters per group. Name the page context. Include a human route.
  4. Day 4: check the knowledge. For each group, ask the assistant ten real questions. Fix gaps in your FAQs, page content or feed. Our guide to organizing FAQs for a chatbot helps here.
  5. Day 5: test the layout. Run the corner-collision check on desktop and mobile.
  6. Days 6 to 7: launch quietly. Turn it on, then read the first conversations.

For question ideas by funnel stage, see our copy-and-paste pre-sales questions.

What to measure after launch

A page-targeted widget is an experiment, and the research on web experimentation says the same: controlled experiments are how companies decide which design changes help. You do not need an elaborate setup. Compare one page group before and after a change, or run two openers on the same page type.

Useful signals for a small site:

  • Open rate per page group. Does the greeting earn a click?
  • Starter-prompt clicks versus typed questions. Tells you if the buttons match real intent.
  • Unanswered questions. Each one is a missing FAQ or a feed gap.
  • Handoff requests. A spike on one page often means the page itself is unclear.
  • Leads captured per page group. Which pages turn conversations into contact details.

Change one thing at a time, and give each change enough traffic to mean something before you judge it.

Common mistakes

  • Targeting by guesswork. Set rules from the questions people ask, not from assumptions.
  • One greeting everywhere. It defeats the point, and visitors learn to ignore it.
  • Interrupting checkout. A pop-up at payment time can cost a sale. Keep it passive or hidden there.
  • No way out. If the assistant cannot help, the visitor needs a person, fast. Interaction patterns for suggested questions, free input and escalation apply here too: offer buttons, allow typing, route to a human.
  • Letting data drift. Page rules do nothing if prices and availability behind them are stale.

How Konvuno fits

Konvuno is an AI website assistant you install with a single script tag and configure from a hosted dashboard, with no developer needed. It answers from your FAQs, website content and product catalog, captures leads into a built-in CRM, and hands off to a human when a question needs one. Shop Connect syncs catalogs from WooCommerce or any Google Merchant or Facebook product feed, so product, price and availability answers stay current. It speaks seven languages, and there is no per-seat pricing.

Plans are published at konvuno.com: Starter at €29 per month, Growth at €79 per month and Business at €199 per month. You can configure and preview before paying. If you want a step-by-step install, read how to add a chatbot to your website without hiring a developer.

FAQ

What is a page targeting chat widget?

It is a chat widget that changes its behavior based on the page the visitor is viewing. The change can be visibility, the opening message, the suggested questions or the content it answers from. Rules usually match a URL, a path or a page category, and a default handles the rest.

Do I need a developer to set up page targeting?

Not necessarily. Many hosted tools let you set rules from a dashboard. Whether a tool offers it, and how detailed the rules are, varies, so check before you commit. The larger job is writing good greetings and keeping the knowledge behind them accurate, which an owner can do.

Should the chat widget appear on checkout pages?

Often it is better hidden or kept minimal there, because checkout is where distractions cost sales. If shoppers regularly ask about delivery or payment at that step, a small help prompt can be worth it. Check your conversation data and test it.

How many page rules should a small site start with?

Four to six groups is plenty: homepage, product or service pages, pricing, help and contact, plus a default. Add a new group only when you see visitors asking distinctly different things on a set of pages.

How do I make visitors understand the assistant knows the page?

Say so in the greeting. Name the product or topic and change the starter prompts to match. Without that, visitors may not realize the assistant can see the page and may paste product names into the chat.

Sources