AI Mode Traffic Hides in Organic Search: How to Estimate It

Short answer: you can't see Google AI Mode traffic directly in Google Analytics, and you can only estimate it. GA4 puts clicks from AI Overviews and AI Mode in the Organic Search channel, in the same bucket as a click on a regular result. Search Console now shows how often your pages appear in those AI features, but only as impressions, with no clicks or queries. The workable method is to measure each landing page's share of impressions that came from AI features, apply it to that page's organic traffic as a low-to-high range, and check the range against two or three independent signals.

This guide is for Shopify and DTC merchants, and their agencies, who track AI traffic and noticed Google's own AI surfaces are missing. Every platform rule below comes from Google's help pages or the Search Central blog, listed under Sources. The estimation methods are ours, with the error bars on each.

AI Overviews

The AI-generated summary Google shows at the top of some regular search results pages, with links to the sources it draws on. Search Console treats an AI Overview as a single position on the results page: every link inside it gets that same position.

AI Mode

Google Search's conversational mode, launched in the U.S. on May 20, 2025. Google describes it as using a "query fan-out" technique that breaks a question into subtopics and searches each one. In Search Console, a follow-up question in AI Mode counts as a new query.

Where does AI Mode traffic show up in GA4?

In Organic Search. Google's default channel group definitions say Organic Search covers visits from non-ad links in organic results, "including Google's AI Overviews and AI Mode". The AI Assistant channel, added in 2026 for visits from assistants such as ChatGPT, Gemini, Deepseek, Copilot and Grok, explicitly excludes them.

That's Google's design, not a tagging bug. A click from an AI Mode answer arrives from Google Search and is classified like any other organic Google click, so no source, medium or channel value tells them apart, and a custom channel group can't either.

Organic Search (GA4 channel)

The GA4 default channel for unpaid search clicks. A session lands here when its source is on Google's list of search sites or its medium is exactly "organic". Google's definition says it includes AI Overviews and AI Mode.

What does Search Console report for AI Overviews and AI Mode?

Two places hold AI data, and they answer different questions.

The standard Performance report (Web search type)

Google's AI features documentation says sites appearing in AI Overviews and AI Mode are included in overall search traffic and reported in the Performance report under the Web search type. The counting rules for those features are:

• Clicks: a click on a link to an external page in an AI Overview or in AI Mode counts as a click. Clicks on links that stay inside Google don't.
• Position: an AI Overview occupies one position, and all its links get that position. AI Mode positions follow the same method as a regular results page.
• Follow-ups: a follow-up question in AI Mode is treated as a new query. Impressions, positions and clicks in the new response are credited to that new query.
• Search Labs: experiments in Search Labs are not included.

So the AI clicks are already inside your Search Console clicks and your GA4 organic sessions. Google's documentation offers no filter in this report that isolates them, and AI Overviews and AI Mode don't appear in the Search appearance list.

The Generative AI performance report

On June 3, 2026, Google announced a dedicated report for visibility in generative AI features, first for a subset of sites. A note on the post says it reached all websites worldwide on August 31, 2026. The help article sets out what it contains:

• Metric: impressions only. An impression is a time links to your site were shown to a user in a generative AI feature on Google Search. The chart counts two results from the same site in one AI feature as a single impression.
• Coverage: AI Overviews and AI Mode, combined in one report. Discover has its own separate report.
• Dimensions: pages (the final URL after redirects, usually the canonical), countries, dates (Pacific Time) and devices.
• Search type filter: "Web: text-based" for typed queries and "Web: multimodal" for image searches such as Google Lens and Circle to Search.
• Limits: the usual Performance report limits apply, including the 1,000-row table cap. Values shown as ~ or - export as zero.

If the report doesn't appear for your property, Google says it is usually because the site hasn't had enough generative AI impressions, or because the site has been excluded from Search generative AI features.

Why not just subtract? Because the Generative AI report gives impressions with no clicks, and Google hasn't published how often AI Overview or AI Mode impressions turn into clicks. An AI Overview can list several sources at one position, and an AI Mode answer may settle the question without a click. So every estimate needs a stated assumption about AI click-through, carried through as a low and a high value.

How do you estimate AI Mode traffic? Five methods and their error bars

No single method is reliable alone. Use the first as your main estimate and two others to check whether it moves the same way.

Method 1: AI impression share by landing page

This is the main estimate. For each landing page, divide its impressions in the Generative AI report by its Web impressions in the standard Performance report for the same dates. That ratio is the page's AI impression share. Then apply it to the page's organic sessions and revenue in GA4, adjusted by a low and a high assumption about AI click-through.

AI impression share

Our working metric: a page's impressions in Search Console's Generative AI performance report divided by its impressions in the Performance report (Web search type), for the same dates, country and device filters. It shows how much of a page's Google visibility came through AI Overviews and AI Mode. It is not a click or traffic share.

1. 01

   Export the Generative AI report by page

   Open the report from Google's help article, set last month's dates, select the Pages tab and export. Keep the search type on "Web: text-based" unless image search matters to you.

2. 02

   Export the Performance report by page

   Same property, same dates, Web search type, Pages tab. Export impressions and clicks. Both tables aggregate by page, so the numbers are comparable.

3. 03

   Join the two on URL

   Match on the page URL in a spreadsheet. Search Console attributes most data to the canonical URL, so join to GA4 landing pages using the canonical form, without query strings.

4. 04

   Compute the share and the range

   For each page: AI impression share × the page's organic sessions × a click-through adjustment. Run it twice, once with a low adjustment and once with a high one. Repeat for organic purchase revenue.

5. 05

   Sum and round

   Add up the low and high results across pages. Report the range rounded, for example "roughly 4–11% of organic revenue", with the method in a footnote.

The adjustment is the judgment call. A high end of 1.0 assumes AI impressions click through as often as ordinary ones, which is generous. A low end such as 0.25 assumes a quarter as many clicks. Google publishes neither, so write your choices down and keep them fixed month to month so the trend means something.

Known errors: the share mixes AI Overviews with AI Mode; the 1,000-row table cap can drop low-traffic pages, so large catalogs should export in sections using a URL filter (for example, product pages and collection pages separately); and generative AI impressions count once per site per feature in the chart but per page in the table. Always use the page-level table, never the chart total.

Method 2: the long-query proxy in Search Console

Because an AI Mode follow-up counts as a new query, conversational questions that reach your pages show up as long queries in the Performance report. Search Console has supported regex filters for queries since April 2021: add a query filter and choose Custom. A pattern such as ^\S+(\s+\S+){7,}$ keeps queries of eight words or more. Track their clicks month over month.

Treat this as a direction check, not a count. People type long queries into ordinary Google Search too, so the proxy overcounts. It also undercounts: Google's documentation says anonymized queries are dropped from totals when a query filter is applied, and long, rare queries are the likeliest to be anonymized. If the long-query clicks and the method 1 estimate rise together, that supports the estimate. If they diverge, find out why before reporting either.

Method 3: landing-page trend comparison

Sort your organic landing pages into two groups using method 1: pages with a high AI impression share and pages with little or none. In GA4, chart organic sessions and purchase revenue for each group over the past six to twelve months. If the high-AI group's organic revenue is growing or shrinking faster than the low-AI group's, AI exposure is plausibly part of the cause.

Seasonality, ranking changes and promotions move groups for reasons unrelated to AI, so compare similar page types (collections with collections) and annotate known promotions and core updates.

Method 4: ask buyers

Add one question to your post-purchase survey: "Where did you first hear about us?" Offer separate answers for "Google search results", "Google's AI answer (AI Overview or AI Mode)" and "ChatGPT or another AI assistant". It's the only method that ties AI exposure to orders, including shoppers who came back later through branded search.

Its errors are recall and labeling: shoppers don't reliably know whether a Google summary was an AI Overview. Read it quarterly, once a few hundred orders have answered.

Method 5: the exclusion holdout (usually not worth it)

Search Console's Search generative AI control, under Settings > Search generative AI for all sites since August 31, 2026, removes a property from AI Overviews, AI Mode and Discover's generative AI features within one to two days, and Google says it isn't used as a ranking signal elsewhere. Switching it off for a month is the cleanest experiment available, but you give up your AI visibility while competitors stay in the answers, and nothing guarantees you're cited as often afterwards. For most stores the cost outweighs a more precise number.

How should AI Mode show up in your monthly AI report?

Keep it beside the AI Assistant numbers, never added into them. The AI Assistant channel (plus your regex group for assistants Google's list misses) is measured: referrers you can count. AI Mode and AI Overviews are estimated. Mixing a measured series with an estimated one hides how uncertain the total is.

• Line 1, measured: AI Assistant and AI-referral sessions, key events and revenue from GA4, as the GA4 AI Assistant channel guide sets up.
• Line 2, estimated: organic revenue plausibly driven by AI Overviews and AI Mode, as a low–high range from method 1, with the click-through assumptions in a footnote.

Read line 2 for its trend, not its level: a range that rises over a quarter on unchanged assumptions tells you something real. Our AI-attributed revenue report template includes an "Organic with AI Mode" caveat row for this. The monthly AI revenue report guide shows how to explain it to a CFO.

Run it on the first working day of each month: export both Search Console reports for the previous month, pull GA4 organic sessions and revenue by landing page, run method 1 with your fixed assumptions, and check the direction against method 2. Then write one sentence, such as "Estimated AI Overview and AI Mode revenue was X–Y, up from A–B on unchanged assumptions; long-query clicks rose in step."

What can't anyone measure yet?

As of October 2026: which of AI Overviews or AI Mode is growing, since the report combines them; which questions put you in the answer, since there are no queries or clicks; and anything automated, since Google's documentation describes no API access. Google says it plans to add metrics over time, so check the help article each quarter.

For the bigger picture of where every kind of AI visit lands (AI Assistant, Referral, Organic and Direct), and what GA4, Shopify and server logs can each see, start with the AI traffic and attribution guide. For how visibility in Google's AI features is earned in the first place, see Google AI Overviews for ecommerce, and for the other AI channels selling your products, the AI shopping channels guide.