How to track AI Overviews and AI Mode traffic: what Google reports, what it hides and a monthly scorecard

The short answer: use the Generative AI report for visibility and your own enquiries for proof

To track AI Overviews and AI Mode, open the new Generative AI performance reports in Google Search Console. They show how often your pages appeared in AI features (impressions), which pages, in which countries and on which devices. They do not show clicks, and GA4 groups the resulting visits under Organic Search, so the honest way to prove value is a monthly scorecard that ties those impressions to enquiries.

Who this is for: a founder, CMO or marketing head who keeps hearing that AI search is taking traffic, wants to see whether it is true for their own site, and needs to decide whether to spend money on it. The decision is simple: do we change our content and budget because of AI search, or not, and what evidence would tell us?

What Google actually launched, and when

On 3 June 2026, Google announced Search Generative AI performance reports in Search Console, with dedicated views for Search and for Discover. The Search view covers generative AI features such as AI Overviews and AI Mode. A note at the top of the same post says that as of 31 August 2026 Google has rolled out these insights to all websites worldwide, so you should not need to be in a test group any more.

Google lists exactly what the reports show:

• Impressions: how often URLs from your site appeared in generative AI features.
• Pages: which URLs appeared.
• Countries: your visibility on a country basis.
• Devices: available for Search results.
• Dates: hourly, daily, weekly and monthly granularity.

Google also states that this data is still included in the overall performance report, so your normal totals have not changed. The new view is an additional, separate lens. The post ends by saying Google is continuing to work with site owners on "additional metrics over time", which is a polite way of saying today's version is limited.

What the report does not tell you

Read the list above again for what is missing. There is no click count, no click-through rate and no position in the dedicated report. Three consequences follow, and they should shape how you present the numbers to your team or board.

1. You can measure being shown, not being chosen. An impression means a URL from your site appeared. It does not mean anyone read it, clicked it or trusted it.
2. You cannot split AI clicks from normal clicks. Google's guidance on AI features and your website says these appearances are reported in the Performance report within the Web search type. They are mixed into your ordinary totals.
3. Do not quote AI traffic as a figure. Any agency, tool or deck that gives you a precise "traffic from AI Overviews" number for Google is estimating. Ask how.

The same Google page says there are no additional technical requirements to appear in AI Overviews or AI Mode: a page must be indexed and eligible to be shown in Google Search with a snippet. So there is nothing to "switch on". The work is the same as before, which is helpful pages that are easy to crawl, a point we cover in our technical SEO basics checklist.

Where GA4 fits, and where it cannot help

Google Analytics Help describes the default channel group, and says Organic Search includes traffic from non-ad links in organic search results, including Google's AI Overviews and AI Mode. In plain terms, a visit that starts from a link inside an AI Overview will usually land in the same bucket as a visit from a blue link. GA4 will not label it as AI.

GA4 can still help in two ways.

1. Watch the landing pages that Search Console says appear in AI features

Take the top pages from the Generative AI report and look at them in GA4 under Reports, Engagement, Landing page, filtered to Organic Search. You are not measuring AI clicks. You are asking a better question: do the pages that Google uses in AI answers also bring engaged visits and enquiries? If a page earns many impressions but few engaged sessions and no enquiries, the page is showing but not persuading.

2. Track other AI assistants separately

Assistants such as ChatGPT and Perplexity usually appear as a referral source, because the visit comes from another site or app. In GA4's Traffic acquisition report, switch the dimension to Session source and look for names like chatgpt.com or perplexity.ai. The sources and volumes change as products change, so check the list yourself rather than relying on a fixed set. Our guide on getting mentioned in ChatGPT and other AI tools explains the visibility side. If you have not set up key events in GA4 yet, start with our GA4 guide for business owners, because none of this matters until a form submission, call or WhatsApp click is counted as a conversion.

The monthly AI search scorecard (copy this)

Most dashboards for this topic fail because they track too much. This is the smallest scorecard that still lets you make a decision. Fill it in once a month, on the same day, from the same reports.

The last row matters more than it looks. Many people read an AI answer, remember your business name and search for it later, or message you directly. That journey shows up as a brand search or a direct visit, not as an AI click. A question on your enquiry form is the cheapest way to see it.

A worked example with clearly labelled assumptions

The numbers below are made up to show the method. They are not benchmarks and not a client result. Imagine a Chennai dental clinic with a service page "dental implants in Anna Nagar" and a blog post on implant costs.

How to read it: impressions more than doubled, but clicks and enquiries barely moved. The post is being used in AI answers, probably because it states a clear cost range with its conditions, but readers get their answer there and do not need to click. The right response is not to celebrate the impressions. It is to ask what the reader still needs after the summary, such as a consultation booking step, a comparison of implant types and a visible WhatsApp button, and to put those near the top of the post. Compliance matters too for clinics: see our guide to NMC advertising rules for clinics before changing claims or prices.

Four decisions the data can support

Decision 1: do we need an AI search budget?

Not before you have evidence. If your scorecard shows few AI impressions, put the money into the fundamentals that feed both classic and AI results: indexed pages, answer-first content, a strong Google Business Profile and reviews. If it shows growing impressions on commercial pages and enquiries are following, then adding depth to those clusters is justified.

Decision 2: which pages to improve first?

Start with pages that have high AI impressions and weak engagement. They already have Google's trust but fail the reader. Add a clear next step, a worked example and fresh, verified facts. Our answer engine optimization guide shows how to write the first lines so that they are quotable.

Decision 3: should we change how we report to the board?

Yes, modestly. Report impressions as visibility, never as traffic. Pair every impression chart with enquiries from organic search. This protects you from a spike that does not pay, and it is the same logic as the layered view in our content marketing ROI scorecard.

Decision 4: when do we stop and re-check?

Set a quarterly review. Google has said additional metrics may come, and the report is new. A dated note in your own file ("clicks not available as of 31 October 2026") saves an awkward debate later.

Set-up checklist for this week

1. Confirm you are a verified owner or full user of your Search Console property, then open Performance and find the Generative AI in Search report.
2. Export the pages list and note the top 10 URLs, the countries and the device split.
3. In GA4, confirm that form submits, calls and WhatsApp clicks are key events. If they are not, fix that first.
4. Create a saved GA4 exploration or report for Organic Search landing pages and add your top 10 URLs.
5. Add "How did you hear about us?" to your enquiry form with options such as Google search, Google AI answer, ChatGPT or another assistant, Instagram, referral, other.
6. Copy the scorecard table into a sheet. Record the first month as the baseline. Do not judge anything yet.
7. Book a 30 minute review on the same date each month. In that meeting, pick one page to improve and one decision to make.

If you track leads in a CRM, tag the source of each enquiry there as well. A simple pipeline view of which sources close is far more useful than traffic charts, and Be12 CRM can hold the source field next to each lead and WhatsApp conversation.

Common mistakes to avoid

• Treating impressions as visits. Impressions show that you appeared. Visits and enquiries show that you helped someone.
• Buying "AI ranking" tools to fill the gap. Third-party trackers sample prompts and estimate. They can be useful for direction, but they are not Google data. Ask how a vendor collects its numbers before you pay.
• Chasing a bot file or special tag. Google says there are no additional requirements for AI features, so be wary of anyone selling one.
• Rewriting everything at once. Change one page, wait a few weeks, and compare. Otherwise you will not know what worked.
• Ignoring the rest of search. Most enquiries for local and service businesses still come from classic results, maps and WhatsApp. Our Chennai local SEO guide covers those.

What to do next

Open Search Console today, find the Generative AI report and record your baseline. Then set up the enquiry source question and the GA4 key events. In 30 days, you will have your first real scorecard, and for the first time a factual answer to the question your board will ask: is AI search sending us business, or only showing our name?

Sources: Google Search Central blog, Introducing Search Generative AI performance reports in Search Console (3 June 2026); Google Search Central, AI features and your website; Google Analytics Help, Default channel group. All checked on 31 October 2026.