Microsoft Advertising Optimization Experiments Are Now Generally Available

Microsoft Advertising says optimization experiments are "now generally available for Search, Shopping, Audience, and Performance Max campaigns". The announcement is part of its September 2026 product roundup, published on 30 September.

This post covers what Microsoft has said about the feature, and the other items in the same roundup that matter if you manage Microsoft Advertising accounts.

What Microsoft announced

According to Microsoft's September 2026 roundup, optimization experiments let advertisers run A/B tests on bidding, targeting and creative elements, with results compared in one place. General availability covers four campaign types:

• Search
• Shopping
• Audience
• Performance Max

The roundup is a summary, so it doesn't spell out the limits that matter in practice: how traffic is split, how long a test must run before results are meaningful, and which settings can be varied per campaign type. Check Microsoft's help documentation for those details before designing a test. Nothing in the announcement says experiments change how campaigns are ranked or priced outside the test itself.

Other items in the same roundup

• HubSpot integration. Microsoft says you can "see which Microsoft Advertising campaigns drive pipeline and revenue, right in HubSpot", with automated lead follow-up. The announcement doesn't say which HubSpot plans or tiers are required.
• LinkedIn company lists. Lists for account-based targeting can now hold up to 10,000 company names, up from 1,000.
• SOAP API retirement delayed. Microsoft says new features and enhancements "will be available in both SOAP and REST until SOAP is deprecated on January 31, 2027". The previous plan was for new features to be REST-only from 1 October 2026. If you maintain an integration built on the SOAP Bing Ads API, you have until 31 January 2027, but new development is better pointed at REST.

What this means

• Test one change at a time. A/B testing is only informative if a single variable differs between the original and the experiment. Changing bids and creative together makes the result impossible to attribute.
• Performance Max experiments are the notable addition. Automated campaign types tend to offer fewer controls, so a supported way to test against a baseline is useful. Whether Microsoft's version is as flexible as its Search testing is something the roundup doesn't say.
• Mind the volume. Microsoft's share of search traffic is smaller than Google's, so low-volume campaigns may take longer to reach a clear result. The announcement gives no minimum.
• Developers have more time, not no time. The extension moves the cutoff for SOAP-only features by months, and the January 2027 date is the one to plan around.

Sources

• Microsoft Advertising: Less busywork, more growth: What's new in Microsoft Advertising this September