Google Ads switches to AI Max: what changes for campaigns

For years we have built Search campaigns thinking that control resided above all in the keywords, queries, ads and landing pages. That control was never absolute, but it was visible: we could indicate where to look and recognize quite well where to intervene. With AI Max, Google moves another part of that work into systems that combine signals, generate texts and choose the most suitable page.

The promise is seductive: intercept searches we hadn't foreseen and adapt the message more quickly. Tension, for me, is not “automation yes or no”. It's understanding which part of the decision we are giving up without first having defined the boundary. If the system expands queries but doesn't know margins, lead quality, or promises the brand doesn't want to make, it can improve a metric and make business worse. Automation doesn't eliminate strategy: it makes it more expensive not to have it.

Time needed

45 minutes for the audit

Difficulty

Intermediate

Cost

It depends on the budget and volume of the campaigns

Tools

Google Ads + Analytics + CRM data

Check

October 23, 2026

What's happening to Google Ads campaigns

Google announced in April 2026 the release of AI Max from beta and a progressive migration of legacy functions. The initial plan included the upgrade of the Dynamic Search Ads, the automatically created assets and the broad match set at campaign level from September.

Then the calendar changed. As Search Engine Land reconstructed, reporting the indications of Google Ads Liaison Ginny Marvin, the automatic migration of DSAs has been postponed to February 2027 to give advertisers more time. The ability to create new DSAs has been restored. The automatically created assets and the broad match at campaign level have instead maintained the transition expected from September 2026.

This distinction is important because many headlines sum it up with “Google eliminates DSAs.” It hasn't happened yet. We are in a phase of differentiated transition, and it is precisely now that it makes sense to build a clean comparison.

How AI Max works for Search

AI Max brings together three areas:

1. search term matching: expands the way your campaign can connect to searches, using broader keywords, ads, page and signals;
2. text customization: adapt titles and messages using available content;
3. final URL expansion: you can choose a different landing page when you consider it more relevant.

Google adds controls for brand, location, text and URL. They are fundamental, but they do not replace the work upstream. A brand audit does not decide which promises are sustainable. A geographic exclusion does not by itself recognize a negative margin zone. A matching system doesn't know that a formally valid lead requires three business hours and rarely closes.

For this reason I would connect AI Max to data that describes value, not just conversion: margin, qualified lead, closed sale, return, customer duration and operating cost.

Simply put: you're no longer just buying a keyword

With the classic model, the advertiser built a relevant part of the map: words, groups, ads, pages. With AI Max provides more materials and boundaries, while the system decides more often which path to use to reach the person.

It's like going from a written itinerary curve by curve to a navigator that recalculates. You get to unexpected destinations more easily, but you need to make sure you know the right destination, the off-limits routes, and the cost you're willing to pay.

The quality of the inputs therefore becomes more important:

• a confusing site offers confusing pages to choose from;
• an incomplete measurement optimizes towards incomplete signals;
• generic assets produce generic combinations;
• an implicit brand strategy doesn't suddenly become explicit because AI enters.

What really changes compared to DSA

Dynamic Search Ads have been using site content for years to intercept searches and generate relevant titles. AI Max preserves this idea, but inserts it into a broader system: it is not limited to the relationship between page and query, it combines signals, creativity and URL selection.

Google describes the transition as a way to find new queries and maintain relevance. I would also read it as a shift in the work unit: from the ad group that we directly control to the system of inputs, rules and outcomes that we must govern.

| First question | Question with AI Max | | --- | --- | | Which keyword is missing? | Which signal is expanding the traffic? | | Which ad has better CTR? | Which combination brings value after the click? | | Which page did I assign? | Which page was chosen and why is it eligible? | | What conversion went up? | What conversion really represents the business? |

The checks I would do before the test

1. Queries and intentions

I would export at least eight weeks of search terms. I would separate useful question, ambiguous question, support query, competition, work, information and purchase. The categories become a basis for exclusions and test reading.

2. Landing pages

I would build a list of allowed URLs, to watch and exclude. I would check titles, promises, availability, speed, tracking and geographical consistency. URL expansion makes the site an active part of the targeting.

3. Brand and text

I would write down what the system can say, what it should avoid, and claims that require proof. I wouldn't leave the policy inside a person's head. I would transform it into rules shared with marketing, legal and commercial.

4. Conversions and value

I would check which actions fuel optimization. A submitted form is not the same as an opportunity; a call is not the same as a sale. Whenever possible, I would return CRM signals and offline value to Google.

5. Experiment and thresholds

I would define first: duration, budget, primary metric, minimum lead quality, tolerable variation in CPA and stop conditions. Without thresholds, every outcome can be recounted as happening after the fact.

What changes for agencies, companies and the advertising market

Agencies will lose value if their service coincides with the manual execution of settings that the platform automates. They will buy if they know how to connect media, data, site, creativity and commercial quality. The work shifts from "moving levers" to designing inputs and verifying consequences.

Companies will have to decide who owns the rules of automation. It cannot be just a media issue: landing pages, feeds, CRM, brand and margins belong to different functions. AI Max makes an organizational problem that already existed more evident.

For Google, the advantage is offering campaigns that adapt to the new way people search, including an increasingly conversational environment. Ask Advisor takes this direction even further: an agent that crosses Ads, Analytics and Marketing Platform and can prepare or suggest actions from a goal expressed in natural language.

Market risk is opacity. If multiple decisions take place in the same system that sells inventory and measures the result, advertisers and professionals must invest more in experiments, proprietary data and independent controls.

How the marketer's daily life changes

There will be fewer hours spent creating manual variations and more hours spent checking materials, queries, pages and signals. Not necessarily less work: different work.

The weekly meeting should not open with "how much did we spend?", but with four questions:

• What did the system learn this week?
• where did you expand queries, messages or pages?
• Which expansion generated real value?
• Which decision should we correct or make more explicit?

This approach connects to the digital strategy with ChatGPT: AI helps sort evidence and hypotheses, but budget, criteria and responsibilities remain human. And it connects to the content factory with human control: automating a production without a proportionate review also amplifies errors.

Three future scenarios for campaigns

Cautious: AI Max as an optional layer

Teams maintain relatively controlled structures and use AI Max on selected campaigns or segments. The platform improves controls and side-by-side testing remains normal.

Likely: The agent becomes the main interface

Ask Advisor and similar tools turn natural language goals into campaigns, analysis, and fixes. The operator controls briefs, permissions, metrics and approval more than individual settings.

Disruptive: advertising and response merge

Ads are increasingly entering conversational experiences and generated responses. Creativity, feed, site reputation and information quality become a single system. The distinction between campaign, content and product data is becoming blurred.

What would I do now

I wouldn't wait until February 2027 to find out how AI Max performs. I would prepare the audit today, choose a campaign with sufficient volume and controllable risk, define thresholds and start an experiment. But I wouldn't migrate everything to obey the direction of the platform. A useful transition maintains a comparison: without baseline, we won't know if AI has improved the business or just changed the way we look at it.

To put this choice within a broader system, I would start from the guide on how to set up a digital strategy with three AI and assign the role of analysis and control to the AI, not the authority over the budget.

Frequently asked questions

Has Google already eliminated Dynamic Search Ads?

No. Automatic DSA migration has been postponed to February 2027. Some other legacy features began transitioning in September 2026.

What activates AI Max?

The main functions are search term matching, text customization and final URL expansion, with controls and settings that must be verified for each campaign.

Is it worth waiting for automatic migration?

No, it is better to use the time for audits and tests. This doesn't mean activating everything immediately: it means arriving at the deadline with your own data and already clear rules.