AEO has a budget now. Teams are rewriting product content, fixing structured data, opening up to AI crawlers and running prompt audits. Then someone in finance asks a fair question: what did we get for it?
Most enterprise eCommerce dashboards can't answer. AI search doesn't behave like the organic channel your reporting was built for. Buyers see your brand in an answer without clicking. When they do click, the visit often arrives without a label. And sometimes the AI answer does its work weeks before the buyer ever lands on your site.
The good news is that measurement improved a lot in 2026. Google added an AI Assistants channel to GA4 in May and dedicated Generative AI performance reports to Search Console in June. Neither tells the whole story on its own. Combined with two things you build yourself, they give you a measurement stack that holds up in a budget conversation.
Why AI search is hard to measure
Three gaps explain almost every "we can't see it" complaint.
- Visibility without clicks. Your page is cited in an AI Overview or a ChatGPT answer, the buyer gets what they needed, and no session is ever recorded.
- Clicks without labels. Visits from AI apps, and links copied out of a chat, often arrive with no referrer. Analytics files them as Direct, which quietly inflates that channel.
- Influence without a visit. In B2B, a buyer may shortlist you from an AI answer, then type your URL a week later or call a rep. The AI tool did the work, and nothing in your analytics points to it.
No single tool closes all three. That's why AEO needs a stack, not a report.
Layer 1: Visibility in Google's AI surfaces
In June 2026, Google launched Generative AI performance reports in Search Console. For the first time you can see how often your URLs appear in AI Overviews, AI Mode and generative features in Discover, separate from standard search results.
Know what it does and doesn't show. At launch it reports impressions by page, country, device and date. It doesn't report clicks, click-through rate or the queries that triggered the appearance. It's a visibility report, not a traffic report.
Bing Webmaster Tools has its own AI performance reporting too, which is worth watching because Bing's index feeds several AI assistants.
How to use it:
- Group pages by template (product, category, guides, blog) and track AI impressions for each monthly. Templates tell you more than individual URLs do.
- Look for pages where AI impressions rise while organic clicks fall. That's the signal that answers are being given without a visit, and those pages need a stronger reason to click: pricing, availability, configurators, comparison tools.
- Note pages with zero AI impressions that should have them. That usually points to a rendering, crawling or structured data problem, not a content one.
Layer 2: AI referral traffic in GA4
For years, clicks from AI assistants landed in GA4's generic Referral channel next to forum links and directory listings. In May 2026, Google added a default AI Assistants channel that catches traffic from the AI tools on Google's own list. It's a real improvement, but it isn't complete: some assistants, Perplexity and Claude among them, can still land in Referral depending on how the visit arrives.
The fix: a custom channel group built on top of the default one.
- In GA4 Admin, go to Data settings, then Channel groups, and copy the default group.
- Add a channel called AI Search, matching session source against a pattern of AI domains, for example: chatgpt.com, chat.openai.com, perplexity.ai, claude.ai, gemini.google.com and copilot.microsoft.com. Add others as they show up in your referral report.
- Match on source, not on medium. ChatGPT adds utm_source=chatgpt.com to many of the links it cites, and those visits can arrive with no medium, or one that isn't "referral". A rule that requires medium = referral misses them.
- Move the AI Search channel above Referral and Organic Search. GA4 assigns each session to the first channel it matches, so an AI channel sitting below Referral never fills.
Two cautions. Don't add your own UTM tags to links you expect AI tools to cite: you'll overwrite the source data you're trying to read. And treat the result as a floor, not a total, because app traffic and copied links still slip through as Direct.
Layer 3: Presence in the answers themselves
Search Console only covers Google. To see how ChatGPT, Perplexity, Gemini and others describe you, you have to ask them, systematically and repeatedly.
A recurring prompt audit covers this layer: a fixed set of 15 to 25 prompts in your buyers' words, run across the major tools on a set schedule, scored for whether you're named, whether you're described accurately, where you rank among competitors, and which sources are cited. We walk through the method in What Does AI Say About Your Brand? Dedicated AI visibility tools can automate the runs once you know which prompts matter.
The metric that matters here is share of voice: across your category prompts, how often you're named compared with your top competitors. It's the closest thing AEO has to rankings, and it moves before traffic does.
Layer 4: Business outcomes
Visibility and visits are leading indicators. Leadership will eventually ask about revenue, and in B2B that means connecting AI exposure to quotes, accounts and orders over long buying cycles.
- Tie the AI channel to conversions. Once AI Search is its own channel in GA4, report its RFQs, account registrations, sample requests and orders next to organic search.
- Carry the source into the CRM. Hidden form fields that capture the session's source and landing page let AI-sourced leads keep that label all the way to closed revenue.
- Ask buyers directly. Add "An AI assistant, such as ChatGPT" as an option in your "How did you hear about us?" field. Self-reported attribution is imperfect, but it's the only way to see influence that never produced a click.
The monthly AEO scorecard
Pulled together, a monthly report needs only a handful of numbers:
- AI impressions in Search Console, by page template
- AI Search sessions in GA4, and the share of total traffic they represent
- Conversions from the AI Search channel: RFQs, accounts, orders
- Share of voice across your category prompts versus your top three competitors
- Accuracy issues found in the prompt audit, and how many were fixed
- The trend in Direct traffic to deep product and category pages, which often hides unlabeled AI visits
Read the trends, not the absolute numbers. Every layer undercounts in its own way, but they undercount consistently, so direction is reliable even when totals aren't.
Common mistakes
- Treating AI traffic volume as the main KPI. AI referrals are usually a small share of sessions. Judge them by what they convert into, not by how many there are.
- Comparing AI and organic click-through rates. AI answers are built to resolve the question on the page. A lower click rate is expected; the question is what the clicks you do get are worth.
- Measuring before fixing access. If AI crawlers are blocked, or your key pages render their content client-side, the reports will just confirm you're invisible. Our guide to faceted navigation and AI crawlers covers the access side.
- Starting without a baseline. Capture all four layers before you change anything, or you'll never be able to show what your AEO work moved.
FAQ
Q: Can GA4 track traffic from ChatGPT? Partly. Visits that arrive with a ChatGPT referrer or its utm_source tag can be captured by GA4's AI Assistants channel or a custom channel group. Visits from the mobile app or from copied links often show up as Direct, so treat your GA4 number as a floor.
Q: Does Search Console show clicks from AI Overviews? Not separately. The Generative AI performance report launched in June 2026 shows impressions by page, country, device and date. Clicks from AI features still count toward your overall Performance report, but they can't be isolated.
Q: What's the best single KPI for AEO? There isn't one. Share of voice in your category prompts is the best leading indicator, and conversions from the AI Search channel are the best lagging one. Report both together.
Q: Is AEO separate from SEO? Mostly no. Google's own 2026 guidance treats optimizing for AI search as part of SEO, and the foundations are the same: crawlable pages, accurate structured data and genuinely useful content. What's different is how you measure it.
The bottom line
You can't manage AEO with the organic search dashboard you built five years ago. Pair Google's new reports with a custom AI channel in GA4, a recurring prompt audit and a direct line to business outcomes. That gives you an AEO story you can defend in a budget meeting, and a clear view of which fixes are working.
If you'd like help setting up the baseline, our SEO + AI Search Visibility Audit covers all four layers: crawler access, AI visibility, analytics configuration and how AI tools describe your brand today.