Ads in AI Overviews and AI Mode: what a shop can control

For a shop in the DACH region there is nothing to book today

You cannot buy these placements, and in the DACH region (Germany, Austria and Switzerland) they so far exist only in part. My argument: what matters is therefore whether you even notice the launch when it comes. As things stand, it will bring no switch and no separate report, and the new inventory lands in the same ROAS average. How hard it hits you depends on your match type mix, because exact and phrase only apply where there is direct purchase intent. Report data from before the launch remains available at daily level for 37 months, according to Google. What you will be missing are defined comparison groups, a launch date and your contribution margin at the time.

Three things are often lumped together: ads alongside AI Overviews, ads within AI Overviews and ads in AI Mode. There is a second mistake as well: the fact that AI Mode is available in a country does not mean ads run in it there.

What runs where: the US and DACH separately

As of November 2026, and this is changing quickly. According to Google's help page, ads alongside AI Overviews are possible wherever AI Overviews exist. That includes DACH, where AI Overviews have been running since March 2025. Ads within the answers run in the US and other markets such as Canada, Australia, India and Singapore, in English only. Google's country list may be incomplete, but no DACH country is on it.

AI Mode itself has also been rolled out in Germany, Austria and Switzerland, but ads are not part of it. In its most recent expansion, Google names Brazil, Mexico and the UK among the next markets, and no DACH country. Google gives no date for DACH.

Which campaigns are eligible, and what does that mean for your mix?

According to Google, existing Search, Shopping and PMax campaigns are eligible automatically. You need neither a new campaign nor a separate budget. The interesting part is the difference within this group: AI-powered targeting, meaning AI Max, broad match, Shopping and PMax, can match contextually to both the query and the AI response. Exact and phrase are only eligible where there is direct purchase intent, not for exploratory questions.

From this I derive an assessment, not a measurement: how strongly your account will be affected at launch depends on your match type mix. A shop that runs almost everything through PMax, Shopping and broad will get a new pattern of queries into the same campaigns and the same budgets. A shop with a large share of exact match in Search will feel less of it in that part, though its Shopping runs along regardless. Exact therefore offers no complete protection, because queries with purchase intent remain eligible.

What you can control and what you cannot

According to Google, three things are outside your control. You can neither book AI Overviews and AI Mode specifically nor exclude them separately from Search. You get no separate report for these placements. And the AI explainers next to your ads are generated by Gemini, partly from public sources such as your website: you cannot write, edit, approve or switch them off.

The existing levers remain under your control:

• Targeting: you decide what share of the budget runs through AI-powered targeting. Judge AI Max and broad match by how they perform today, not by some future inventory.
• Exclusions: according to Google, negative keywords and brand exclusions continue to apply. Well-maintained lists are less work before the launch than after it.
• Budget and target: bids and budget go through the same auction with no surcharge. Your target ROAS applies on average, so weak inventory can run along beneath it.
• Your website: because the explainers are also built from public sources, I advise keeping shipping, returns, prices and product information on the site free of contradictions. How much effect this has is not proven, but it costs little and helps the shop anyway.

Why the launch would barely show in ROAS

A worked example with invented, round figures. Suppose a shop spends €20,000 a month and generates €100,000 in revenue, ROAS 5.0, with a 40% contribution margin before ad spend. Its break-even ROAS is therefore 2.5. After a launch, €2,000 moves into exploratory queries that only bring ROAS 2.0.

In the report, ROAS falls by 0.3, which easily passes as a monthly fluctuation. Contribution margin falls by 12%. According to Google, the figures are only included in aggregate in campaign reports, and the same applies to the search terms report: you cannot tell which click came from an AI answer.

The example only holds under its assumption. If the new inventory mainly brings new customers whose repeat purchases carry the low ROAS, the same drop would be good news. Telling the two apart requires a baseline.

Which baseline you should set now

You can still retrieve the report data later, but you should define the groups beforehand: per country, exact and phrase on one side, broad, AI Max, PMax and Shopping on the other. Your ads already run alongside AI Overviews in DACH, so that is already contained in the before figure.

1. Share of long search terms: cost and revenue from search terms of five or more words and from questions. The cut-off is my working definition, not a Google rule. The search terms report provides this, as far as it shows the terms. After the launch it is only a rough signal, CPC and contribution margin tell you more.
2. CPC per group and country.
3. Contribution margin after ad spend per group, not just ROAS.
4. Share of new customers per group, as far as your tracking allows.

Write down today what will be hard to reconstruct later: margin, share of new customers from the shop system, conversion setup and campaign structure. When news of the launch in your country arrives, note the date and compare before and after, ideally against the same period in the previous year, because otherwise seasonality distorts the result. If the two groups diverge, that points to the new inventory, but it is no proof, because exact and phrase can appear there too. If one DACH country launches before the others, the rest can serve as a rough comparison. If you want real evidence of incremental revenue, you need an incrementality test.

For small accounts, monthly figures per group and country often do not work, because there are too few conversions. In that case use quarters and a single country.

With US business you will see it sooner

If you already sell to the US, you can observe the effect earlier in English-language traffic. It only transfers to a limited extent, because language, product range and competition differ. The same goes for US studies. A Similarweb analysis evaluated by Aleyda Solis shows, for headphones, the paid share of clicks rising from 16% to 36% year on year. That points to a shift in a few US categories. It does not demonstrate an effect of ads in AI answers, and it says nothing about DACH.

If your account runs almost entirely on brand and exact terms, the effort is probably greater than the insight. I have found no independent measurement of whether ads in AI answers bring retailers additional profit. I would not shift any budget for this today, nor restructure any campaign. Recording groups and margin levels costs little, and without them the launch disappears into the average.