Google Ads in AI Search Results

The quarterly planning meeting is on Thursday and you have to defend a search budget. Somebody has already forwarded the article: AI answers are eating clicks, the ten blue links are finished, why are we still funding keywords. Fair question. So you open Google Ads to see what is happening in your own account, and there is nothing there to look at. No AI Overviews column. No placement filter. Spend, clicks and conversions sit in one blended line that looks roughly like last quarter. You cannot tell whether the thing everyone is writing about has reached you at all.

Short answer: Google reports ads inside AI Overviews as top ads with no separate breakdown, so you cannot isolate that placement in your account. Treat it as a measurement problem: baseline your search terms by intent type now, split branded from generic, and judge the channel on cost per outcome as the query mix shifts.

The takeaways

• The placement exists and your account will not show it to you. Ads in AI Overviews are reported as top ads with no separate breakout, so there is nothing to segment, target or exclude.
• This is a measurement problem, not a forecasting one. Nobody can tell you how far the shift goes, but you can baseline your own query mix by intent type in about 30 minutes.
• Impressions are the first number to stop trusting. Split branded from generic, then read conversion rate and cost per outcome separately for each intent type.

Can you see AI Overview ads in your Google Ads account?

No. Ads that appear in AI Overviews are reported as top ads in Google Ads, without separate breakout reporting, as Sharp Innovations' February 2026 rundown of the placement puts it. There is no segment to pull, no filter, no column. The same write-up lists what you cannot do: target only AI Overview placements, or opt out of AI Overview eligibility. AdControlCenter's June 2026 read of early AI Mode testing adds that there is no placement-level bid modifier for it either. So the placement is live above your ad, it is spending your budget, and Google Ads hands you one blended number that quietly contains it. Worth sitting with for a second: every strategy post you have read about this was written from outside your account, from SERP scrapers and platform announcements and other people's aggregates. That matters before you let one of them talk you into a restructure.

What changes when an AI answer sits above your ad?

The query mix moves before your metrics do. An AI answer is good at closing out an informational question: what is this, how does it work, which type do I need. Someone who gets that answered in place has no reason to click anything, paid or organic. A person typing an emergency repair, a price comparison or a brand plus "buy" is in a different position. They want a destination, and an answer box is not one. Mapping queries to where the buyer stands is the same exercise as matching keywords to awareness stage, and it is the exercise that survives this shift. So the first thing to move is the composition of what reaches you. Informational impressions thin out. Commercial ones hold, or get pricier because more advertisers are now competing for a smaller set of clickable moments. Read that in aggregate and the two movements average into one flat line.

Which numbers move first, and which ones lie?

Split branded from generic before you read any trend at all. Branded search is largely insulated from an answer box, so leaving it in your totals props the whole account up while the generic half erodes underneath. Once they are apart, impressions become the least useful number on the screen. They fall when the mix shifts, and a fall on its own is neither good nor bad news. Conversion rate per intent type is the honest read. If informational impressions drop while your conversion rate climbs and cost per sale holds, you lost traffic that was never going to buy, which is a saving. If cost per sale climbs alongside it, the auction tightened and that becomes a budget decision rather than a strategy crisis. Keep an eye on impression share, but treat it as a hint: the page it measures a share of no longer looks the way it used to.

How do you baseline this before it reaches you?

Export your search terms report now, while the ground is still recognisable. Take the top 50 queries by spend and tag each one by intent type: informational, comparison, transactional, branded. Half an hour of work, and it buys you a before picture that nobody can sell you later. Repeat it monthly and watch the proportions rather than the raw counts. Here is the part the strategy posts leave out. The instrument is partly blind to the thing you are measuring: AdControlCenter's testing notes that AI Mode queries, being conversational and long-form, may not appear in search terms reports with the same fidelity as ordinary search. You are measuring the residue, what still reaches your account, and not the answer box itself. That bind is familiar from measuring ChatGPT ads, where the platform cannot count its own influence. Own evidence with a known gap still beats a forecast about somebody else's account.

Should you restructure the account now?

Not on this evidence. The prescription doing the rounds is to collapse 50 exact-match ad groups into a handful of broad themes and hand the rest to Smart Bidding. That may turn out to be right. It is also a large, slow, learning-resetting bet placed on a shift you have just established you cannot see directly. Make the small reversible moves first: tighten negatives, keep the intent split in your reporting, and let cost per outcome tell you when something has changed. The discipline underneath is the one that stops a buyer blaming a creative for a bad market week. Adscalr's ad intelligence layer reads fatigue from a multi-day slope and runs a market-wide check beside it, so a move that hit everyone at once does not get scored as your ad breaking. The environment shifting and your work failing are different findings, and this quarter it is worth being able to tell them apart.