How to Measure AI Visibility: Metrics, Prompts and Reporting for Ecommerce Brands
A practical framework for measuring how ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews talk about your products: prompt sets, the six metrics that matter, referral tracking, and a weekly reporting cadence.
To measure AI visibility, build a fixed set of real buyer prompts for each priority product (category, comparison, attribute, occasion, problem and brand questions), run them across each AI engine on a regular schedule, and record whether your product is mentioned, recommended, described accurately and cited. Report mention rate, recommendation rate, accuracy and share of voice per engine and per SKU, then connect them to referral traffic from AI sources.
- Measure per product and per engine. A single brand-level number hides where you're losing.
- Use real buyer prompt types: category, comparison, attribute, occasion, problem and brand reputation.
- Track six metrics: mention rate, recommendation rate, position, accuracy, share of voice and citations.
- AI answers vary, so sample repeatedly and report trends, not single screenshots.
- Connect visibility to outcomes using AI referral traffic, such as ChatGPT's utm_source=chatgpt.com tag.
On this page
"Are we showing up in ChatGPT?" is now a common question in brand meetings. The usual answer is a screenshot someone took last Tuesday. That isn't measurement. It's one sample of a system that gives different answers to different people, phrasings and engines.
You can't manage what you can't measure, and AI visibility can be measured if you set it up properly. Here's a framework that works whether you track manually in a spreadsheet or use a dedicated platform.
Why is AI visibility harder to measure than SEO rankings?
Classic SEO has a stable unit: a keyword, a ranking position and a click-through rate. AI answers break all three:
- No fixed ranking. An answer might name three products, one product or none, in prose rather than a list.
- Personalization. OpenAI says ChatGPT considers context such as Memory and custom instructions when choosing products, so two shoppers can get different answers.
- Many engines. ChatGPT, Gemini, Perplexity, Copilot, Google AI Overviews and marketplace assistants each draw on different sources.
- Fewer clicks to count. Many shoppers read the answer and never visit a site, so traffic alone understates your influence.
The fix is the same one survey researchers use: a fixed sample, repeated on a schedule, and trend lines instead of anecdotes.
Which prompts should you track?
Start with the questions real buyers ask, not your keyword list. XENA's SearchPanel classifies buyer prompts into types that make a useful template:
| Prompt type | What it looks like | Why it matters |
|---|---|---|
| Category | "best protein powder for beginners" | Sets the shortlist when the shopper has no brand in mind |
| Comparison | "Brand A vs Brand B for sensitive skin" | The engine picks a winner, and the loser may go unnamed |
| Attribute | "vegan collagen without added sugar" | Spec-driven; complete, machine-readable listings decide who qualifies |
| Occasion | "gift for a new runner under $50" | High intent; surfaces specific products |
| Problem | "how to stop my dog's paws cracking" | The engine decides if a product is the answer at all |
| Brand reputation | "is [your brand] legit?" | Shapes trust at the final step |
| Market variants | The same question in another country | Availability, pricing and local sources change who gets named |
For each priority SKU, write several prompts of each relevant type, in the phrasing shoppers actually use. Pull language from reviews, customer service tickets, search terms reports and Q&A sections.
What metrics should you report?
Six metrics cover most of what matters:
- Mention rate. The share of relevant prompts where your product is named at all.
- Recommendation rate. The share where it's positively recommended, not just listed or mentioned with a caveat.
- Position. Where you appear when several products are named (first, second, later).
- Accuracy. Whether the engine describes your product correctly: price, size, ingredients, compatibility. Wrong facts are a fixable data problem and can cost sales even when you're mentioned.
- Share of voice. Your mentions as a share of all brand mentions in your category's prompt set, compared with named competitors.
- Citations. Which URLs the engine cites as sources: your site, your marketplace listing, a retailer, a review site or a forum. This tells you where to fix things.
Report each metric per engine and per SKU. A brand-wide average can look healthy while your best seller is invisible in the engine your buyers use most.
“How often does AI mention you vs. competitors for the same query? XENA tracks citation frequency, average ranking position, and share of voice across hundreds of real buyer prompts per product.”
How do you turn the metrics into a score?
A composite score makes trends easy to communicate. XENA's SearchPanel gives every product a 0–100 AI Visibility Score reflecting how consistently engines mention, recommend and correctly describe it compared with competitors. It uses four bands:
| Score | Band | What it means (per XENA) |
|---|---|---|
| 0–39 | Critical | AI engines rarely mention you; competitors dominate relevant queries |
| 40–64 | Developing | Occasional mentions; content gaps and missing signals are holding you back |
| 65–79 | Competitive | Appearing regularly; targeted optimization needed to reach the top |
| 80–100 | Dominant | AI consistently recommends you first |
If you build your own score, document the weighting and keep it fixed. A score that changes definition every quarter can't show progress.
How do you connect AI visibility to revenue?
Visibility is a leading indicator. To connect it to outcomes, layer in traffic and conversion data:
- ChatGPT referrals. OpenAI says ChatGPT automatically adds
utm_source=chatgpt.comto referral URLs. Filter for it in Google Analytics. - Other AI referrers. Check referrer data for domains like perplexity.ai, gemini.google.com and copilot.microsoft.com.
- Google AI features. Google says AI Overviews and AI Mode traffic is counted in Search Console under the "Web" search type. It isn't broken out separately, so watch trends on pages you've optimized.
- Marketplaces. Amazon doesn't report assistant-driven sessions separately in a simple way, so track conversion and sales trends on SKUs whose listings you've improved, against a comparable control group.
Google also notes that clicks from AI Overviews tend to be higher quality, with visitors more likely to spend time on the site. On its own site, SearchPanel reports +32% higher conversion on AI referral traffic.
What does a weekly AI visibility report look like?
Keep it to one page:
- Headline score or mention rate per engine, with week-over-week change.
- Top 5 SKUs by visibility gain and top 5 by loss.
- New competitor appearances in your core prompts.
- Accuracy issues found: wrong price, outdated specs or a missing variant, with an owner for each fix.
- Top cited sources this week and any new third-party sites.
- AI referral sessions and conversions vs. last week.
- Changes shipped (listing edits, schema fixes, content) so you can link causes to effects.
What mistakes should you avoid?
- Testing while signed in with your own history. Personalization skews results. Use clean sessions or a tool that controls for it.
- One prompt, one run. Answers vary, so run each prompt several times and record the rate, not a yes or no.
- Ignoring accuracy. Being mentioned with the wrong price or old formula can hurt more than not being mentioned.
- Measuring without acting. Tie every gap to a fix. Our 90-day AEO plan turns measurement into a weekly work queue.
New to the topic? Start with What Is AEO?, then come back to set up your baseline.
Frequently asked questions
What is an AI visibility score?
A single number summarizing how consistently AI engines mention, recommend and correctly describe a product compared with competitors. XENA's SearchPanel uses a 0–100 score with four bands: Critical (0–39), Developing (40–64), Competitive (65–79) and Dominant (80–100).
How many prompts do I need to track?
Enough to cover each prompt type for each priority product, with several phrasings per question. A small manual program might start with 20–30 prompts per hero SKU. Automated tools track far more; XENA's SearchPanel reports tracking 2,847 buyer prompts per brand daily.
Can I see AI Overviews traffic in Google Search Console?
Yes, but not separately. Google says traffic from AI features is included in Search Console's Performance report under the Web search type, alongside regular search results.
How do I track traffic from ChatGPT?
OpenAI says ChatGPT automatically adds utm_source=chatgpt.com to referral URLs, so you can filter for it in Google Analytics or similar tools. Also check referrer data for AI domains such as chatgpt.com and perplexity.ai.
How often should I measure?
Weekly is a practical cadence for most brands: frequent enough to see the effect of changes, slow enough to avoid overreacting to normal answer variation.
Find out which of your products AI recommends.
Score a product across ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews with a free SearchPanel audit. Or book a meeting and we'll walk through a live AI Visibility Score with you.
Sources
- XENA SearchPanel: AI Visibility Score
- SearchPanel: Prompt tracking
- Google Search Central: AI features and your website
- OpenAI Help Center: Publishers and Developers FAQ
- OpenAI Help Center: Shopping with ChatGPT Search
XENA results quoted in this article are as published by XENA Intelligence on the linked pages. Third-party facts are from the linked official documentation, checked Oct 26, 2026.
