Tracking
AEO tracking is how you find out what ChatGPT, Claude, Gemini and Perplexity tell buyers about your products, and whether it changes after you fix something. It is close to what tools call AI search monitoring or AI brand monitoring, with one difference that matters for sales: you track the questions buyers ask before they buy, product by product, and you read whether the answer got your facts right. This guide covers what to track, how to set it up, how often to run it, and what the numbers can’t tell you.
Last updated October 9, 2026 · ~6 min read
The short answer: pick a fixed set of real buyer questions, ask them to the AI assistants your buyers use, keep the full answer text with the date, and record for each product whether it was mentioned, whether your pages were cited and whether the facts were right. Repeat on a schedule and after every page change, and compare like with like.
| Signal | What it tells you | Watch out for |
|---|---|---|
| Mention | Your brand or product name appears in the answer | A mention is a text match. The answer may advise against you |
| Citation | The answer links to or lists one of your pages as a source | A marketplace listing or reseller page may be cited instead of yours |
| Facts stated | Whether the price, size, compatibility, warranty or other facts in the answer are correct | Wrong facts usually trace back to a page that is missing, outdated or inconsistent |
| Who else is named | Which alternatives the assistant puts next to you for that question | It changes between runs; look at the pattern, not one answer |
| AI referral visits | Visits from AI assistants to your site, in Google Analytics 4 | Many AI answers lead to no click at all, so visits undercount influence |
Illustrative example, not customer data.
| Buyer question | What the answer said | What it means |
|---|---|---|
| “Can I take a 20,000 mAh power bank on a plane?” | Named two competitors; your model not mentioned | Your page doesn’t state the watt-hour rating, so the assistant can’t confirm it qualifies |
| “Will this dishwasher fit a 24-inch opening?” | Mentioned your model, cited a retailer listing | The retailer states the dimensions more clearly than your product page |
| “Does this sensor work with a Modbus controller?” | Mentioned your model with the wrong protocol | An old spec sheet still lists the previous version |
Answers can take weeks to change after a page update, and no one controls how an assistant decides. Judge the trend over several rounds.
Tracking is only worth the effort if a gap turns into a change. When an answer leaves you out or gets a fact wrong, find the fact the question needs, check whether your product page, listings and spec sheet state it with a source, fix it where buyers and assistants read it, and ask the question again. Each gap is a missed buying opportunity you can see. If you are comparing tools for this, see what AI visibility tools measure.
Hasmord includes AI answer checks: it asks the questions your buyers ask through the Anthropic, Google, OpenAI and Perplexity APIs and keeps each answer, with mentions, citations and recommendations counted separately. It doesn’t stop at tracking: it shows which product facts are missing, conflicting or unconfirmed on your pages, drafts the fixes your team approves, and asks the same questions again after a change. AI referral visits come from Google Analytics 4; inquiries, deals and contracts feed back into the same loop (coming in v3.0).
AEO tracking means asking AI assistants the questions your buyers ask, on a schedule, and keeping a record of each answer: whether your product was mentioned, whether your pages were cited, and whether the facts stated about it were right. It shows how AI answers about your products change over time and after you change your pages.
They overlap. AI brand monitoring and AI search monitoring usually mean tracking how often a brand is named across many prompts. AEO tracking for products goes one level down: the buyer questions behind a purchase, product by product, and whether the answer got the facts right.
Track a fixed set of core buyer questions every week or two, and run them again after every change to a product page. Answers vary from run to run, so judge the trend over several rounds rather than a single answer.
Not in the same way. Model APIs such as those from OpenAI, Anthropic, Google and Perplexity can be asked directly. AI Overviews appear on Google results pages and can only be observed there, and Copilot can’t be queried this way. Ask any tool which surfaces it queries directly and which it does not cover.
Related: AI visibility tools · measuring AEO’s impact on sales · sales questions to AI answers · product evidence · AEO vs SEO.