Google is giving advertisers a little more room to experiment before handing bigger decisions over to AI.

The company is expanding AI Max for Search campaigns with new testing and planning features aimed at answering a fairly practical question: what happens if an advertiser changes the budget, adjusts an ROI target, or lets AI Max handle more of a Search campaign?

Instead of making the change and waiting to see whether it works, advertisers will have more ways to test the impact first.

The additions include multi-campaign A/B testing, support for brand and location controls in AI Max experiments, and expanded Performance Planner forecasting.

Google Brings Multi-Campaign A/B Testing to AI Max

One of the bigger changes arrives in September.

Advertisers will be able to test different budgets and ROI targets across several Search campaigns through a single A/B experiment. Until now, testing AI-driven changes could become awkward when advertisers wanted to understand the effect across more than one campaign.

This pushes the experiment beyond a narrow campaign-level question.

An advertiser could, for example, test whether increasing budgets across a group of campaigns actually produces enough additional business to justify the extra spending. The same approach can be used when experimenting with different return-on-investment targets.

For agencies and advertisers running large Google Ads accounts, that matters. Performance rarely exists inside a neat single-campaign box.

AI Max Experiments Keep Brand and Location Controls

Google is also addressing something that could have stopped some advertisers from testing AI Max in the first place: control.

AI Max experiments will support brand and location controls while a test is running.

That means advertisers won’t necessarily have to loosen important targeting restrictions simply to find out whether AI Max improves performance. A company that only wants campaigns operating within particular geographic areas, for instance, can maintain those boundaries during the experiment.

The same principle applies to brands with strict advertising requirements.

It sounds like a small adjustment. In practice, it could make AI Max testing far more realistic for companies that cannot simply give Google’s automation unrestricted freedom.

Performance Planner Gets More Useful Before Campaign Changes Go Live

Google’s Performance Planner is getting a bigger role as well.

Advertisers will be able to forecast how adjustments to bidding or budget targets could affect the performance of existing campaigns before making those changes.

If Google’s forecast looks worthwhile, suggested changes can then be applied directly to campaigns with a single click.

That creates a much shorter path between planning and execution.

Forecast the change. Look at the expected impact. Decide whether the numbers make sense. Apply it.

Of course, a forecast is still a forecast. Actual campaign performance can move for plenty of reasons that a planning tool cannot perfectly anticipate.

But having another layer between “maybe we should increase the budget” and actually spending the money isn’t a bad thing.

Google Wants Advertisers to Test AI Before Fully Committing

The broader direction here is interesting.

Google has been pushing more automation into advertising for years, but AI Max moves that strategy deeper into Search campaign management.

AI Max uses Google AI for functions including search-term matching and asset optimization. Google’s existing AI Max experiments already allow advertisers to split traffic within an existing Search campaign between a control with AI Max disabled and a trial where AI Max is active. Google says this approach can produce faster insights while reducing setup and synchronization problems associated with maintaining separate campaign copies.

The new tools don’t slow Google’s automation push. They make it easier to measure.

That’s an important distinction.

Advertisers aren’t just being given another AI switch to turn on. Google is building more infrastructure around the switch so marketers can test what happens before making broader changes.

Why These AI Max Updates Matter for Advertisers

Multi-campaign experiments could be especially useful for larger advertisers because budget and ROI decisions are often made across groups of campaigns rather than one campaign at a time.

Brand and location controls tackle another problem. Businesses with strict geographic, compliance, franchise, or brand requirements may be more willing to experiment with AI Max if they don’t have to abandon those restrictions during testing.

Then there’s Performance Planner.

Combining forecasting with one-click implementation nudges Google Ads toward a workflow where AI doesn’t merely optimize campaigns after decisions are made. It increasingly becomes part of deciding what those changes should be.

That’s where this update becomes more significant than another collection of Google Ads buttons.

It also fits a broader shift in digital marketing, where AI search and discovery are beginning to reshape how brands think about visibility beyond traditional search rankings.

AI Is Becoming Part of the Google Ads Decision Layer

Google’s advertising automation used to be relatively easy to spot. Automated bidding handled bids. Responsive ads mixed creative assets. Smart campaigns simplified setup.

The boundaries aren’t quite as tidy anymore.

AI Max combines automation across targeting, matching and creative optimization, while these latest tools extend AI-assisted decision-making into experimentation and planning.

Google is essentially building a feedback loop: plan a change, forecast it, test it, measure the results and potentially deploy it more broadly.

Humans still decide whether those recommendations make business sense.

That part hasn’t disappeared.

But the machinery surrounding that decision is becoming increasingly automated.

For advertisers, the useful question probably isn’t whether Google Ads will use more AI. That direction is already obvious.

The question is whether Google’s new testing tools provide enough evidence to know when the AI is actually helping.

Sources