Google is adding three new testing and forecasting features to AI Max for Search campaigns. The one that matters most shows up in September 2026, and it lets you run a single A/B test across multiple Search campaigns instead of testing one campaign at a time. The other two are rolling out now, including the ability to run an AI Max experiment without turning off your brand and location controls.
What Google Announced
All three of these build on the one-click AI Max experiments that already exist inside Google Ads. Here is what is new.
- Cross-campaign A/B testing, coming in September. You will be able to run one A/B test across multiple Search campaigns at the same time and see how budget and ROI target changes affect results across the account, not just inside a single campaign.
- AI Max experiments with brand and location parameters, available now. You can run an AI Max experiment while keeping your brand and location controls turned on. Before this, testing AI Max meant loosening the guardrails you put in place on purpose.
- An expanded Performance Planner. The Performance Planner can now forecast how a change to your bidding strategy or budget targets would affect an existing campaign, and you can apply Google's suggested change in one click.
Why Testing Across Multiple Campaigns Matters
Single-campaign experiments have always had the same blind spot. You raise the ROI target on one campaign, that campaign looks better, and you call it a win. What you cannot see is what happened to everything else in the account.
One of the biggest mistakes I see is treating a campaign like it lives in its own world. It does not. When one campaign gets more aggressive, it pulls impressions and clicks that another campaign was competing for, and it pulls spend that was funding something else.
A cross-campaign test is the only way to know whether you actually added revenue or just moved it around. That is a much better question to be answering.
Keep in mind, this only works if your budgets are not fighting each other to begin with. If two campaigns are limited by the same daily spend, the test result will not tell you much on its own, so it is worth understanding how budget pacing works before you start.
Brand And Location Controls Remove The Main Objection
This is the update I am most interested in, because it removes the reason a lot of advertisers never tested AI Max at all.
Plenty of businesses cannot serve outside a specific area. Plenty of brands cannot afford to have their name matched against queries nobody approved. Up until now, running a clean AI Max experiment meant giving up some of that control just to get a readable result, and that is a trade most people were not willing to make.
Being able to keep brand and location parameters on during the experiment means the test finally reflects how you would actually run the campaign. That makes a huge difference in whether you can trust the result.
What To Do Before You Test
The brand and location controls are live now and cross-campaign testing lands in September. Either way, the work that makes a test worth running happens before you start it. Here is the process.
- Pick your test campaigns now. Choose two or three Search campaigns with enough conversion volume to give you a real answer. A campaign getting four conversions a month is not going to tell you anything.
- Write down your baseline. Record cost per conversion, conversion value, and ROAS for each campaign over the last 30 and 90 days before you touch anything.
- Check your conversion tracking first. Every one of these features depends on Google reading accurate conversion data. If your conversion tracking is counting button clicks or duplicate form loads as conversions, the experiment will happily optimize toward the wrong outcome.
- Decide what a win looks like. Pick the number that decides the test before the test starts, whether that is cost per lead, revenue, or return on ad spend. Deciding afterward is how you talk yourself into a bad result.
One more thing on the Performance Planner. A forecast is a projection built on your own historical data, not a promise. I would treat the one-click apply button as a starting point you verify rather than an answer, and I would make sure you know what each bidding strategy is actually optimizing for before you switch.
Question to Answer:
Which two or three Search campaigns will you test AI Max on first, and what specific number tells you the test worked?
In Summary
Cross-campaign testing is the real story here. Being able to test budget and ROI target changes across several Search campaigns at once gets you much closer to the question you actually care about, which is whether a change grew the business or just shuffled results between campaigns.
Use the time before it lands in September to clean up your conversion tracking and write down your baselines. A testing feature is only as good as the data underneath it. If you want the bigger picture on how these pieces fit together, my Google Ads optimization guide covers the account structure this kind of testing depends on.
0 comments