The Mentality Shift Behind AI Max

AI Max is easy to misread as another toggle in campaign settings. I believe that reading is too narrow.

AI Max is not a new campaign type. It is a layer on Search campaigns whose job is to make those campaigns flexible enough for AI-shaped results: AI Overviews, AI Mode and conversational search where the user needs a much more flexible experience than before.

What I am interested in is what AI Max and automatic campaigns represent for Google and for the panorama where paid media and AI search share the same demand.

What AI Max is actually for

For AI Max to work, Google needs the account to give up two old certainties:

1. Matching gets fuzzier. Keywords become signals. Conversations get compressed into thematic or synthetic keywords.
2. Ads and URLs get flexible. Text and landing pages adapt so the response can sit inside the conversation, not only next to a static Exact match.

That is the mentality shift. Keywords stop being the hard target of the account. They become one intent signal among several: creatives, site, brand lists, exclusions, conversions.

If you still manage these campaigns with a 2019 Exact-match brain, AI Max will feel broken when, in reality, is doing a different job.

What automatic campaigns represent for Google

The future problem Google is trying to solve is to figure out how do ads stay relevant when demand looks like a dialogue?

When users ask long questions, refine them mid-conversation, or get answers inside AI surfaces, a campaign built only on fixed keywords, fixed text and fixed URLs is a bad fit. Automatic layers (AI Max on Search, the wider family of keywordless and generative behaviours, Performance Max) are how Google keeps paid inventory relevant in that world.

The panorama Google wants is clear: flexible campaigns for conversational demand. Automatic campaigns are how paid stays in the room when search stops looking like a keyword box.

What this represents for paid media in an AI world

For us, the risk is answering a structural shift with a checklist.

The useful framing is dual:

• Open: let matching, text and URLs flex enough to enter conversational and AI-shaped demand.
• Close: define brand, messaging and URL boundaries so the account still represents the business.

If you only open, you optimise for Google’s inventory problem. If you only close, you stay out of the experiment and loose market positioning.

AI Max is designed to step outside what you built. That is how it finds conversational queries. The same design means it does not know your legal limits, which pages sell nothing, or that inventing a price to win CTR is unacceptable. Automatic does not mean unsupervised. It means your control model has to move from keyword lists to a perimeter.

That dual job is the strategy and secondary features are where you implement it.

Why engaging with automation makes sense even when results are not always great

Google’s published success stories frame AI Max as an upside. Industry data is less tidy: ROAS results swing from positive to negative, as usual, depending heavily on the business, the account and other factors.

But I still want practitioners fluent in this system, because:

1. Direction beats comfort. If a slice of demand is moving into conversations and AI surfaces, learning flexible campaigns now is cheaper than inventing a control model under pressure later.
2. You can practice with a small blast radius. A deliberate pilot with a perimeter beats discovering automatic layers only when they are already shaping delivery.

Of course we should not trust the case studies blindly, we should get good at paid inside an AI search world while we still choose the pace.

The perimeter automatic campaigns need

Flexibility in AI Max stands on three legs.

Brand inclusions and exclusions

Tell the system which brands you will show for, and which you will not. Exclusions at campaign level. Inclusions at campaign or ad group. In case of conflict, the exclusion wins.

There is also the larger brand searches control: everything relevant, your own lists, or non-branded only.

Automatic matching loves brand traffic: yours and competitors’. That inflates generic campaigns, cannibalises brand campaigns, and wrecks incrementality. Brand control is how measurement stays honest when search term matching widens.