Microsoft Advertising has opened a beta for Microsoft Performance Max experiments, giving advertisers two structured ways to test the campaign type that has resisted controlled testing since launch. The news surfaced on July 3, 2026, after Microsoft’s Ads Liaison Navah Hopkins confirmed the beta on LinkedIn, and PPC News Feed founder Hana Kobzova spotted the help documentation days earlier.
Until now, experiments in Microsoft Advertising were limited to Search campaigns. Those have been renamed Search optimization experiments, and two new experiment types now sit alongside them under Campaigns > Experiments: the uplift experiment and the upgrade experiment.
The beta is in limited availability. If the options do not appear in your account, Microsoft has directed advertisers to request access through support or their account representative.
The uplift experiment measures incrementality
The uplift experiment answers the question every advertiser running Performance Max eventually asks: is this campaign generating conversions that would not have happened anyway, or is it taking credit for demand my other campaigns already capture.
The setup is straightforward. You select a Performance Max campaign as the treatment and a group of comparable non-PMax campaigns as the control. The control group runs your existing mix as usual. The treatment group runs the same mix plus the Performance Max campaign. The difference between the two groups is your measured uplift.
Microsoft’s help documentation recommends that the comparable campaigns share overlapping audiences, conversion goals, and landing pages with the Performance Max campaign being tested. The minimum duration is two weeks, with roughly eight weeks recommended for results that hold up statistically, and Microsoft advises excluding the first week of data because early results are noisy while the system learns. The metrics tracked are conversions, conversion value, CPA, and ROAS.
The upgrade experiment tests the switch before you commit
The upgrade experiment addresses a different decision: should you move an existing campaign to Performance Max at all.
You pick an existing campaign as the control and a Performance Max version as the treatment, then split traffic between them at a percentage you choose. Microsoft recommends a 50/50 split for the most reliable comparison. Duration guidance matches the uplift experiment: two weeks minimum, around eight weeks recommended, first week excluded from analysis. Confidence intervals are adjustable between 80% and 95%.
One operational detail worth noting: several settings lock once an experiment starts. For upgrade experiments, that includes the control and treatment selections, the start date, and the split. For uplift experiments, the treatment campaign, the comparable campaign set, and the start date lock in. Plan the setup carefully, because you cannot adjust it mid-flight.
Why Microsoft is shipping this now
The timing is not random. In a May 7, 2026 blog post, Microsoft claimed that advertisers using its Performance Max see an average 8% increase in incremental conversions, citing internal data from September 2024 to September 2025. At its Activate 2026 partner event on May 19, the company added customer acquisition goals focused on incrementality, new exclusion and targeting controls, and previewed this experiments framework.
Read together, the message is clear: Microsoft knows incrementality skepticism is the main blocker to PMax adoption, and it is now handing advertisers the tools to test the claim themselves. That is a meaningful shift. Vendor-reported averages are easy to dismiss. A controlled test in your own account is not.
It also closes an obvious gap with Google Ads, where campaign experiments have existed for years. Microsoft has been steadily porting Google’s automation playbook, and we covered how AI took over PPC bidding and placement in our breakdown of AI in PPC. Testing infrastructure was the missing piece.
What this means for B2B SaaS advertisers
Performance Max has a specific reputation problem in B2B: it reports conversions enthusiastically while offering limited visibility into where budget actually went. The core worry is cannibalization, where PMax intercepts branded and high-intent search demand that cheaper campaigns would have converted anyway. We examined that exact failure mode in our analysis of whether Performance Max cannibalizes search, and the uplift experiment is the first native Microsoft tool built to answer it with data instead of opinion.
For teams already running Microsoft Ads alongside Google, the practical sequence looks like this. First, check Campaigns > Experiments for access, and request it if the options are missing. Second, if you already run PMax on Microsoft, start with an uplift experiment: it will tell you whether the campaign deserves its budget. Third, if you have been holding off on PMax, use an upgrade experiment on one mid-priority campaign rather than converting your account on faith.
Budget the timeline honestly. With an eight week recommended duration plus a discarded first week, you are looking at roughly two months before you have a defensible answer. Structure your test before you start, because the locked settings mean a sloppy setup wastes the entire window. If your account structure is not clean enough to define comparable control campaigns, fix that first. Our Google Ads audit checklist covers the same structural hygiene that applies on Microsoft.
Microsoft’s later October 2026 advertising updates also introduced additional Performance Max targeting controls and a planned search term landing-page report. Those improvements can help explain campaign behavior, while experiments remain the way to test incrementality.
The bigger picture: incrementality testing is becoming table stakes across ad platforms, and advertisers who build the habit now will make sharper budget calls than those who keep trusting platform-reported conversions at face value. We dug into why PMax eats budgets without that discipline in our Performance Max whitepaper for SaaS.
At OneMetrik, we run this kind of controlled testing as standard practice across Google Ads management and Microsoft Ads for B2B SaaS clients. If PMax is a growing share of your spend and you cannot prove what it adds, an uplift experiment is the cheapest insurance you will buy this quarter.
Frequently asked questions
What are Microsoft Performance Max experiments?
They are two new beta experiment types in Microsoft Advertising: an uplift experiment that measures the incremental conversions a Performance Max campaign adds over a control group, and an upgrade experiment that splits traffic between an existing campaign and a Performance Max version before a full transition.
How do I get access to the beta?
Check Campaigns > Experiments in your Microsoft Advertising account. Availability is limited during the beta, and Microsoft has directed advertisers without access to request it through support or their account representative.
How long should a Performance Max experiment run?
Microsoft’s documentation sets a two week minimum and recommends around eight weeks for statistically reliable results, with the first week of data excluded from analysis because early performance is noisy while the system calibrates.
Which metrics do the experiments measure?
Both experiment types track conversions, conversion value, CPA, and ROAS. Confidence intervals are adjustable between 80% and 95%, and core settings such as campaign selections, start date, and traffic split lock once the experiment begins.