Google expands AI Max with new A/B testing and a performance planning tool for Search budgets

Google is giving advertisers more ways to stress-test and scale performance for Search campaigns built with AI Max, the company said in a product update posted Wednesday.

The latest additions, which are scheduled to begin rolling out in September, extend Google’s “one-click experiments” approach by allowing marketers to test different budgets and return-on-investment (ROI) targets across multiple Search campaigns within a single A/B test. The change is designed to show how scaling up campaign spend and adjusting ROI goals may affect results at the campaign level, not just for isolated experiments.

In the update, Google also said that advertisers who use specific controls, such as brand and location settings, will be able to run A/B tests while keeping those guardrails enabled. The company positioned the new functionality as a way to validate the impact of AI Max without overriding the brand or geographic constraints some advertisers treat as non-negotiable.

Beyond experimentation, Google introduced “Performance Planner,” a planning feature intended to help advertisers anticipate how adjustments to levers like bidding or budget targets could influence existing campaign performance. Rather than focusing solely on what an A/B test might reveal after the fact, the feature is framed as a forward-looking tool that connects planned changes to expected outcomes.

Google described Performance Planner as one-click capable, with the ability to apply suggested changes directly to campaigns. That is a notable workflow shift for many advertisers who traditionally move more slowly from planning spreadsheets into production settings, and who may want tighter feedback loops when using AI-driven bidding and budget optimization.

Taken together, the changes appear to address two recurring questions in performance marketing with automated systems: how to test whether bigger budgets and different ROI targets are actually improving outcomes, and how to make those decisions without losing brand and location restrictions that keep campaigns aligned with business and compliance requirements.

AI Max, which Google mentions as the foundation for the new tools, is part of its broader suite of machine-learning-driven advertising automation for Search. In practical terms, it aims to optimize campaign performance using algorithms that continually adjust toward advertiser objectives, while still requiring marketers to define where the automation should operate and how success should be measured.

For many mid-market and enterprise advertisers, the operational problem is not simply whether AI can optimize bids or budgets, but whether the optimization can be evaluated in a way that matches how companies manage risk. Running experiments that span multiple campaigns can reduce the time spent coordinating separate tests, while preserving the ability to compare outcomes under controlled conditions.

Google’s post did not provide specific performance metrics, case study results, or details about statistical methodologies, nor did it spell out exactly how long experiments will run or what reporting will be available within the interface. It also did not outline whether every campaign type or configuration is eligible for the new multi-campaign A/B test approach, or what limitations might apply for advertisers with complex account structures.

Still, the message is clear about intent: Google wants advertisers to use AI Max while preserving testing discipline, and to convert planning into action more quickly through suggestions that can be pushed to campaigns in one click. The company is also directing customers to attend Rethink 2026, a Google event, described in the post as a way to learn industry best practices for scaling performance with Google AI.

For advertisers watching rollout timelines, the primary next step is to see how the September release fits into existing A/B testing workflows and whether it supports the brand and location controls their accounts rely on. In the near term, companies are likely to focus on whether Performance Planner’s recommended changes are transparent enough to be trusted internally and whether the one-click application reduces the lag between planning and optimization.