Google adds multi campaign experiments and new planning tools to AI Max
Google announced three new capabilities for AI Max for Search campaigns: testing budgets and ROI targets across multiple Search campaigns in one A/B test, running experiments with brand and location controls enabled, and previewing the likely impact of changes in Performance Planner before applying them in one click.
Three additions on the measurement side
Per an announcement published on the Google ads and commerce blog on August 20, 2026, AI Max for Search campaigns is gaining new testing and planning capabilities. The post is signed by Brandon Ervin, Director of Product Management for Search Ads.
Search Engine Land confirmed the news the same day and described the package as an expansion that lets advertisers see the impact of changes before committing to them.
There are three headlines: multi campaign A/B testing, support for brand and location controls inside experiments, and a new forecasting and apply flow in Performance Planner.
What each capability does and when it lands
First, multi campaign A/B testing. Advertisers will be able to test different budgets and return on investment targets across multiple Search campaigns inside a single A/B test. Google frames the purpose as seeing how scaling campaigns affects the bottom line. This capability rolls out in September.
Second, brand and location controls in experiments. AI Max experiments can now run with those settings enabled. In Google framing this lets an advertiser evaluate AI Max without removing their own guardrails.
Third, Performance Planner. The tool now shows how changes such as bidding or budget targets may affect existing campaign performance, and suggested changes can be applied to campaigns in one click.
Per the Search Engine Land assessment, multi campaign experiments give a clearer view of what happens to overall business performance when campaigns scale, while control support makes AI Max testing more practical for businesses with strict requirements.
The announcement shares no performance statistics and no customer examples. Brand and location controls and the Performance Planner update are available as of the announcement, while multi campaign testing waits for September.
As automation grows, the burden of proof grows with it
The commentary in this section is ours. The direction on ad platforms in recent years is clear: decisions keep moving to the algorithm, and what stays with the advertiser is targets, signals and budget. In that setup the most valuable asset is not a control panel but the ability to produce proof. Stronger experiment tooling means exactly that.
The real value of multi campaign testing is that single campaign experiments could not answer the question we keep asking. Loosen the target in one campaign and that campaign grows, but does the whole account grow, or is the budget simply moving over from the neighbouring campaign? When several campaigns sit inside the same experiment, that question becomes measurable.
Being able to experiment with brand and location controls enabled closes a quiet but important gap. Running that kind of test previously often meant switching a guardrail off and letting the platform run free, which made the test unacceptable in the first place. A business with brand protection in place can now evaluate the system with the guardrail still on.
The one click apply in Performance Planner cuts both ways. It adds speed, but it can also weaken the habit of asking how reliable the forecast is. The easier a suggestion is to apply, the less it gets examined. That risk is our assessment.
Directionally this sits on the same line as other measurement moves of recent weeks; the change in Branded Searches measurement also redefined what the report actually shows.
What it means for businesses in Türkiye
The assessment below is not in the sources, it is our reading. The announcement states no Türkiye specific timeline or restriction.
In small and mid sized ad accounts here, the pattern we see most often is deciding without experimenting. Budget goes up, results look better, the increase becomes permanent. Meanwhile season, campaign period and competitor activity are all moving at once. Stronger experiment infrastructure is a good reason to change that habit.
The second point is budget scale. On small budgets an A/B test can end before it reaches a statistically meaningful result. Multi campaign testing can help here, because data spreads across a wider set rather than a single campaign. Even so, the test has to reach enough data before any decision.
The third point is brand protection. In many businesses here, searches on the brand name carry a large share of conversions. Experimenting with brand controls enabled makes it possible to evaluate automation without putting that traffic at risk.
Three practical steps. Verify that your conversion measurement works correctly before entering an experiment. Build the test on a campaign set that represents the business outcome rather than the whole account. Read the result together with revenue and profitability, not with a single metric.
The UNALSOFT take
On the advertising management side our way of working does not change: the more automated the platform gets, the more proof matters. New experiment tooling is not a new strategy for us, it is the existing discipline made easier to apply. Before scaling budget in a client account the question is always the same: does this increase grow total sales, or are we buying the same sales at a higher price? Multi campaign experiments make that answer calculable. We stay cautious about one click suggestions though, because applying a suggestion is easy and undoing its effect is not always.
Measure before you scale the budget.
Let us look together at whether a meaningful experiment can be built in your account.