Microsoft Advertising makes optimisation experiments generally available

What changed

Microsoft confirmed on 30 September that optimisation experiments are generally available for Search, Shopping, Audience and Performance Max campaigns. Each experiment splits budget and traffic between a control and a treatment, and a winning change can be applied to the original campaign.

General availability resolves earlier uncertainty

Week 39 treated the capability as documentation with an unresolved launch date. Microsoft’s 30 September post supplies the missing confirmation of general availability. That date relates to Microsoft’s confirmation, rather than necessarily identifying the first day any account could see the function.

This is stronger evidence than an isolated account control or undated help page. It still does not establish that every campaign will meet all eligibility conditions or that experiments guarantee an actionable result.

Keep the first hypothesis narrow

Microsoft says advertisers can test four bid or creative changes within the same traffic. The capacity to include several changes does not remove the need for interpretable design. If unrelated bid and creative changes are bundled, the treatment may win without revealing which decision should be repeated elsewhere.

Editorial suggestion: begin with one business hypothesis, one primary success measure and one principal difference between control and treatment. Record guardrails such as cost, lead quality or revenue where relevant. Use the broader four-change capacity only when the combined package itself is the decision being tested.

From experiment to implementation

Before applying a winning treatment, check whether promotions, seasonality, tracking changes or other external conditions shifted during the run. Confirm that the measured outcome matches the campaign’s business objective rather than relying on a convenient proxy alone.

The ability to apply a result to the original campaign is a workflow feature, not proof that performance will persist after the test. Preserve the experiment dates, settings and result so the later campaign can be compared with the conditions under which the treatment won.

Where the result is inconclusive, do not force a winner merely because the tool permits application. Revisit the hypothesis, detectable difference and stability of the test conditions.

Why it matters for advertisers

General availability gives Microsoft advertisers a structured way to compare bid or creative changes without relying only on before-and-after trends, provided the test remains interpretable.