Google is giving advertisers something AI Max badly needed: a cleaner way to prove whether more automation deserves more budget.
On August 20, 2026, Google announced new Google AI Max testing tools that will let advertisers test different budgets and ROI targets across multiple Search campaigns in a single A/B test. The multi-campaign testing capability is expected to roll out in September.
Google is also expanding AI Max experiments so advertisers can test with specific brand and location controls enabled. At the same time, Performance Planner can now model bidding and budget changes and make it easier to apply suggested changes directly to campaigns.
That sounds like three separate product updates.
For paid media teams, it is really one bigger change: Google is reducing the distance between testing, forecasting and changing live spend.
What Google announced
The update adds three important pieces to Google’s Search automation workflow.
| Update | What changes | Why marketers should care |
|---|---|---|
| Multi-campaign A/B tests | Test different budgets and ROI targets across multiple Search campaigns in one experiment | Scaling decisions can be tested instead of relying only on forecasts |
| More AI Max experiment controls | Keep specific brand and location controls enabled during tests | Teams can test automation without removing important campaign guardrails |
| Faster Performance Planner execution | Model bidding and budget changes and apply suggested changes more directly | Planning becomes faster, but so does the path from recommendation to live spend |
Google already offered one-click AI Max experiments that let advertisers compare a control with an AI Max treatment inside Search campaigns. The new announcement extends that testing mindset further into budget allocation and ROI targets.
Timing also matters.
Google is gradually moving more Search campaign functionality into AI Max. Automatically Created Assets and campaign-level broad match configurations are still scheduled to begin auto-upgrading to AI Max in September 2026.
However, Google updated its timeline for Dynamic Search Ads on June 11. The automatic DSA migration will now begin in February 2027, giving advertisers additional time to prepare.
Our earlier breakdown of the AI Max for Search campaigns transition explains the broader shift and should now be read alongside Google’s updated migration timeline.
The latest Google AI Max testing tools give advertisers a better way to answer the next question: if AI Max performs, how far should the account scale?
The bigger shift is from feature testing to budget testing
Testing AI Max itself is useful.
Testing what happens when you give the system another 20%, 30% or 50% of budget is more commercially interesting.
Google says advertisers will be able to test different budgets and ROI targets across multiple Search campaigns in a single A/B test. That moves Google Ads A/B testing closer to the decision finance and growth leaders actually care about.
The question is no longer simply: “Does AI Max produce more conversions?”
It becomes: “If we spend more while maintaining a defined return target, does the additional spend still produce qualified pipeline at an acceptable cost?”
That distinction is especially important after Google’s recent changes to target-based bidding and budget-limited campaigns. Google is giving its systems more responsibility for finding volume around the targets advertisers provide.
That means the quality of the target matters more than ever.
A campaign optimizing toward qualified pipeline gives the system a meaningful business objective. A campaign optimizing toward every demo form submission gives it a much easier objective, including students, competitors and low-intent leads your sales team may never contact. No experiment fixes a bad conversion definition.
Why the new Google AI Max testing tools matter for marketers
For B2B SaaS marketing, four parts of this update deserve particular attention.
1. Scaling decisions can have a cleaner control
One of the worst ways to evaluate paid search performance is to increase budgets across several campaigns, wait a few weeks and then try to determine what caused the change.
Budget changed. Auction conditions changed. Search volume moved. Competitors adjusted spend. Sales follow-up may have changed.
Suddenly everyone has a different explanation for the result. Multi-campaign testing gives marketers a better chance of isolating the impact of a scaling decision before treating a forecast as evidence.
This fits the direction Google outlined at Google Marketing Live 2026, where AI Max was part of a much broader shift toward automated Search execution, AI-powered bidding and more revenue-aware optimization.
The opportunity is not simply to scale faster. It is to test whether scaling deserves to happen.
2. Brand and location controls remove a real testing objection
Automation experiments are less useful if running the test requires removing the controls that made the original campaign commercially sensible.
Expanded AI Max for Search campaigns experiments can retain specific brand and location controls during testing.
For a SaaS advertiser separating branded from non-branded acquisition, protecting geographic sales territories or excluding regions without sales coverage, these are not minor settings.
They determine what is actually being tested. If an experiment removes those guardrails, the advertiser may not be testing AI Max at all. They may be testing an entirely different campaign strategy.
Teams reviewing these settings should also revisit their Google Ads campaign structure before scaling. Automation becomes harder to evaluate when brand demand, competitor demand and high-intent generic searches are mixed into the same measurement bucket.
3. Performance Planner is becoming an execution tool
Performance Planner has traditionally helped advertisers estimate how changes to spending, bidding and other campaign variables could affect future performance. The workflow is becoming more actionable.
Google says advertisers can model bidding and budget changes and apply suggested adjustments more directly to campaigns. That saves time.
It also removes some of the friction that previously existed between receiving a recommendation and turning it into tomorrow’s budget.
Our operating rule would be simple: Forecast automatically. Approve manually.
Google’s models can use campaign and auction data to estimate potential performance. They do not understand every commercial factor surrounding the campaign.
They do not know whether sales capacity is constrained next month, whether a board-level CAC target changed yesterday or whether the last ten conversions were actually qualified opportunities.
That is why our broader guide to when to trust Google Ads automation and when to override it still applies.
Faster execution makes human review more important, not less.
4. Reporting quality becomes the bottleneck
Google keeps shortening the amount of time required to analyze, forecast and modify campaigns.
Its recent Ask Advisor reporting update lets advertisers investigate performance using natural-language questions.
The new Google AI Max testing tools move in the same direction on the execution side by making experiments, forecasts and budget decisions easier to run.
Eventually, the limiting factor becomes the data underneath them.
If your primary conversion is a form fill, Google’s systems can optimize, forecast and test against form fills very efficiently.
That still does not make those form fills pipeline.
For B2B advertisers, better automation makes offline conversion imports, CRM stages and qualified revenue signals increasingly important.
The AI Max workflow before and after this update
| Area | Earlier workflow | New direction |
|---|---|---|
| AI Max feature testing | Test AI Max inside a Search campaign | Test AI Max while preserving more campaign controls |
| Budget scaling | Increase spend and compare performance afterwards | Test different budgets across multiple campaigns |
| ROI targets | Adjust targets and monitor what happens | Compare alternative ROI targets through experiments |
| Brand and location guardrails | Controls required separate consideration during testing | Specific controls can remain active in the experiment |
| Planning | Forecast possible performance changes | Forecast and apply suggested changes more directly |
The important point is not that automation has suddenly become trustworthy. It has become easier to test. That is a much more useful improvement.
How to test the new Google AI Max tools in September
The September rollout creates a good reason to review Google Ads budget planning before simply enabling another experiment.
A practical testing sequence would look like this:
- Record the baseline first. Capture spend, qualified conversions, cost per qualified lead, pipeline value and the current CPA or ROAS target before starting the experiment.
- Write one clear hypothesis. If you change both budget and an ROI target, document exactly what you expect to happen. Otherwise, the experiment may produce a result without telling you which decision actually drove it.
- Keep business-critical controls enabled. Use brand and location settings when those restrictions reflect your real GTM strategy.
- Measure the experiment downstream. For lead-generation campaigns, compare SQLs, opportunities and pipeline, not only Google Ads conversions. That should be standard practice for Search campaign optimization.
- Review before clicking Apply. Treat a Performance Planner recommendation as an input to a decision, not the decision itself.
Teams should also continue inspecting search terms throughout the experiment.
AI Max expands Google’s ability to match Search traffic based on intent and other signals. Understanding how keyword match types and query expansion work therefore remains useful even as individual keyword-level controls become less central to campaign management.
What should B2B SaaS teams measure?
The easiest mistake is to run a technically valid experiment against a commercially weak KPI.
For a B2B SaaS company, the experiment should ideally be evaluated at several levels:
| Measurement layer | Example metric |
|---|---|
| Google Ads | CPA, ROAS, conversion rate |
| Lead quality | MQL rate, SQL rate |
| Sales pipeline | Opportunities created, pipeline value |
| Revenue | Closed-won revenue, CAC |
| Efficiency | Cost per SQL, cost per opportunity |
AI Max may increase conversion volume while reducing the proportion of qualified leads.
It may also increase CPA while creating substantially more pipeline.
Neither outcome can be judged correctly from the Google Ads conversion column alone.
That is why the quality of your conversion tracking increasingly determines the quality of your automation.
OneMetrik Takeaway
The most useful part of Google’s August 20 announcement is not another AI feature.
It is that Google is giving advertisers a more disciplined way to test whether AI-driven scale actually deserves more money.
That is progress.
But the Google AI Max testing tools still optimize and test around the business signals advertisers give them. Better experiments cannot rescue poor conversion tracking, mixed campaign intent or a target CPA that has little relationship with pipeline economics.
At OneMetrik, we would treat the September release as permission to test more carefully, not permission to scale faster.
Establish the baseline. Preserve the important controls. Measure qualified pipeline. Then make the person managing the account explain why the “Apply suggested changes” button deserves to be clicked.
If you are preparing for the September AI Max changes, a structured Google Ads audit is a sensible place to start. The question is no longer whether Google can automate more of Search. It is whether your account gives that automation the right target.