AI Overviews and AI Mode Ads: What You Can Actually Measure

Four similar names, four different questions

Google's AI Overview ad documentation is specific: ads inside Overviews are included in Top Ads, with no separate breakdown for that placement. Advertisers also cannot target only the in-overview placement or opt out of it. Those are documented AI Overview rules, not automatically rules for every AI Mode format.

For AI Mode, Google's May 2026 announcement describes tests of Conversational Discovery ads and Highlighted Answers. It does not supply an isolated performance report for those formats. Do not take the narrower AI Overview FAQ and cite it as though it documents both experiences.

AI Max reporting answers a different question: how campaign automation matched searches, created text or selected destinations. Google's reporting guide includes search-term/headline/landing-page combinations. A row in that report is not proof that the ad appeared in AI Mode.

Organic visibility is separate again. Search Console's generative AI performance report documents impressions of site links in AI Overviews and AI Mode. That report is impressions-only: it does not provide clicks or position. Its impressions are not paid ad spend, clicks attributed to an ad campaign or ad ROAS.

An aggregate result cannot reveal the missing split

Suppose a Search campaign spent $20,000 and recorded $100,000 in purchase value. Its reported ROAS is 5×. You also have a screenshot of one ad inside an AI Overview and a rise in Top Ads impressions.

Neither observation tells you how much of the $20,000 served inside Overviews or how much of the $100,000 belongs to that placement. Assigning 10% of spend to it because it appeared in a handful of screenshots would manufacture the missing denominator.

Use the result you can support: the campaign reported 5× ROAS, and an observed ad appeared inside an Overview. The screenshot establishes that instance, not the placement's total delivery or contribution.

The same restraint applies to an improvement after a feature launch. A rising conversion total can coincide with a promotion, a budget change, a different query mix or more completed conversions from older clicks. The timing alone does not show that an AI search experience caused the increase.

This does not make the campaign impossible to manage. It changes which decisions its reports can justify. You can evaluate the campaign's commercial result and fix a poor destination without inventing a separate return for a placement you cannot isolate.

Use the available detail to answer a real question

If lead quality deteriorated, start with the leads and the searches behind them. Check whether the destination promises the right service and whether the conversion still represents a useful enquiry. The lead-quality guide explains how to follow those outcomes beyond a form submission.

For an AI Max campaign, inspect the dedicated search-term and landing-page view when the question involves what people searched, the headline they saw and where they landed. Google's guide warns that filtering the ordinary search terms report to the AI Max match type leaves out Other search terms. Do not mistake the visible filtered rows for a reconciled total.

Search text itself has limits. Google's ad-prioritization documentation says complex search experiences can be represented by an approximation of intent. A conversational-looking phrase is therefore not a reliable placement detector.

If the question is whether enabling AI Max improved the campaign, use the AI Max experiment guide. A campaign experiment tests that feature under its design and assumptions. It still does not automatically isolate the effect of appearing inside AI Mode.

Do not pull every available report by habit. Inspect the part of the account that can resolve the decision, and identify the point where the available evidence ends.