- In AI Mode, Google breaks a question down into many related searches. Ads show up because they match the intent, not because a keyword matches word for word.
- For ads in AI answers, Google requires AI-powered targeting: broad match keywords, AI Max, Performance Max or Shopping campaigns.
- In Germany, ads currently appear above or below the AI Overviews. Google is testing new formats directly in AI Mode in the US.
- Without clean conversion tracking and a well-maintained product feed, the automation optimizes for the wrong signals.
Google Search is changing fundamentally. Instead of matching search terms to pages, Google answers more and more questions itself: in AI Overviews above the search results and in AI Mode, which has been available in German since fall 2025. For advertisers, it is the biggest shift in years, because intent recognition is replacing exact keyword matching.
Advertisers who stick with old account structures lose visibility in these answers. Those who adapt their campaigns reach customers while they are still researching. In this article, we show how ads are served in AI search, where they appear in Germany today, and how to adapt your campaigns step by step without unnecessary risk.
In AI Mode, the intent behind the question is what counts
Classic Google Ads was straightforward: a user searches for “buy running shoes”, you bid on that keyword, your ad shows.
In AI Mode, Google doesn’t wait for an exact phrase. When a user asks a complex question, such as “Which running shoes are right for my first marathon?”, the AI breaks it down behind the scenes into many related searches. Google calls this query fan-out:
- “best marathon running shoes beginners”
- “running shoes cushioning test”
- “marathon training gear”
Your ad can appear because your offer matches the user’s intent, even if they never typed your keywords. Ads within an AI answer have to fit both the question and the content of the answer. That lets you reach potential customers earlier, while they are still researching rather than just before they buy.
In Germany, ads currently appear around the AI answers
Where ads appear in AI search depends on country and language. As of October 2026:
| Placement | Availability | Campaign types |
|---|---|---|
| Above or below AI Overviews | in all markets with AI Overviews, including Germany | Search, Shopping, Performance Max, App |
| Within AI Overviews | in English, in selected countries such as the US, Canada and Australia | Search, Shopping, Performance Max |
| New formats in AI Mode | being tested in the US | Built on AI Max and Performance Max |
In May 2026, Google unveiled new formats for AI Mode. Conversational Discovery Ads answer a user’s specific question with ad copy that Gemini writes for that exact search. Highlighted Answers show relevant ads within AI Mode’s recommendation lists. Over the coming months, AI-powered Shopping ads are set to follow, along with a sales assistant for lead generation that answers prospects’ questions directly in the ad, based on your website.
There are two limitations to be aware of: there is no dedicated report for ads in AI Overviews yet; they show up in reporting as top-of-page ads. And you can neither target nor exclude this placement specifically. In sensitive industries such as finance, health or gambling, Google does not show ads within AI Overviews.
AI-powered targeting is a prerequisite for ads in AI answers
For placements in AI answers, Google relies on AI-powered targeting. That includes broad match keywords in Search campaigns, AI Max for Search campaigns, Performance Max and Shopping campaigns. Exact match keywords on their own barely cover the many ways a question can be phrased.
Broad match on its own is risky, though. Without guardrails, budget flows into irrelevant clicks.
Three components keep broad match in check
For intent-based targeting to work in your favor, three components have to work together:
- Broad match keywords to cover the full breadth of a topic.
- Smart Bidding with target CPA or target ROAS. Only then can the AI tell which of the many searches actually drive revenue and profit.
- Negative keywords as your most important lever. Because broad match casts a wide net, you have to actively exclude whatever doesn’t fit.
Think in intent clusters rather than keyword lists. Instead of 50 variations of “running shoes”, 10 to 15 broad match terms covering purchase, problem-solving (such as knee pain when running) and information are often enough.
AI Max replaces Dynamic Search Ads
The easiest way into AI placements for Search campaigns is AI Max. It adds search term matching, text customization and final URL expansion to existing Search campaigns, with controls for brand, location and text guidelines.
Google is automatically migrating Dynamic Search Ads to AI Max by February 2027, and has been migrating automatically created assets and the campaign-level broad match setting since September 2026. With all AI Max features, Google sees on average 7% more conversions or conversion value at a similar CPA or ROAS, compared with search term matching alone. Test the switch with an experiment first. That way you compare the results before switching the entire campaign. You then see which search terms and landing pages AI Max chooses before it affects the whole budget.
Without clean data, the AI optimizes for the wrong goals
Ads in AI Mode are only as good as the data you feed the automation. Two areas determine success.
Conversion tracking tells the AI what success looks like
The AI has to learn what matters to your business. Basic tracking isn’t enough. You need:
- Conversion values: pass real basket values so the AI can tell a €5 customer from a €500 one. Values that reflect your margin are even better.
- Enhanced conversions: they capture conversions even when browsers block cookies.
- Micro-conversions: track pre-purchase steps such as “Add to cart” as secondary conversions. They give the AI extra signals during the learning phase without skewing bids.
Our guide to conversion tracking explains the full setup.
Your product feed is the raw material for Shopping ads
In e-commerce, ads in AI search are built directly from your product data. An incomplete feed costs you these placements.
- Title: use all the space you have. Instead of “Nike Pegasus”, write “Nike Air Zoom Pegasus 40 men’s running shoes neutral marathon black”.
- Images: high-resolution product images on a clean background are a must.
- Freshness: prices and availability have to be synced continuously via the API.
Performance Max runs on strong assets
Alongside AI Max, Performance Max is one of the campaign types that get you into AI search. It combines Search, Shopping, YouTube, Discover, Gmail and Maps in a single campaign.
Success depends less on settings than on assets. Give the AI enough material to work with: many images rather than one, and several headlines that speak to different reasons to buy:
- Rational arguments: price, delivery time, shipping.
- Emotional arguments: how running feels, the freedom of being outdoors.
- Proof points: reviews, bestseller status.
Also use audience signals, such as your existing customers and website visitors. They don’t restrict delivery, but they help the AI find the right new customers faster. For how Performance Max and AI Max differ, see our comparison of the two campaign types.
We transition your campaigns in three phases
The most common mistake is switching everything overnight. Move all campaigns to broad match and Performance Max at once, and you often lose control. That is why we work in three phases.
Phase 1: Lay the foundations (weeks 1 to 2)
Before any additional budget goes in, we get tracking and the feed in order. We also build an extensive negative keyword list of at least 100 terms to protect the launch of broad match keywords.
Phase 2: Test and learn (weeks 3 to 5)
We launch a test campaign or experiment with around 20 to 30% of the budget, using your most important keywords as broad match or adding AI Max.
The campaign runs for 14 days without any bid changes. The AI needs this time to learn. Each day, we only review the search terms and exclude anything irrelevant.
Phase 3: Scale up (from week 6)
If the contribution to profit is right, we raise the budget gradually, for example by 10% a week, and slowly wind down the old campaigns.
Conclusion: In AI search, the quality of your data is decisive
AI Mode is changing how people search on Google, and with it, how ads get found. In Germany, ads still appear around the AI answers for now. But the new formats in the US show where things are heading. Switch your campaigns now, and you gather data and learn your customers’ intents before these formats roll out widely.
The key is the quality of what you feed the AI: creative no one else has, complete data and rigorously maintained exclusions. For how your company can also show up in organic AI answers, see our article on Generative Engine Optimization.
