Microsoft Advertising has expanded A/B testing for Search, Shopping, Audience, and Performance Max

Microsoft Advertising Expands Optimization Experiments: A/B Testing for Search, Shopping, Audience, and Performance Max

On September 30, 2026, Microsoft Advertising announced the general availability of Optimization Experiments for four advertising campaign types: Search, Shopping, Audience, and Performance Max.

This tool allows advertisers to conduct controlled A/B tests, compare the results of different settings, and implement successful changes without having to evaluate their effectiveness solely on a "before and after" basis.

What is Microsoft Advertising Optimization Experiments?

Optimization Experiments is a built-in Microsoft Advertising tool for testing changes to advertising campaigns through controlled experiments.

It works by comparing two variants:

Control — the original campaign with existing settings.

Treatment — an experimental version of the campaign in which the advertiser changes a specific parameter.

Microsoft Advertising distributes advertising traffic and budget between the variants and then compares their performance.

For example, an advertiser can test whether a new bidding strategy actually reduces CPA. Instead of changing the settings of an existing campaign, they create an experimental version with the new strategy and compare results over the same time period.

This approach helps isolate the impact of the tested change from seasonality, demand fluctuations, and other external factors.

What parameters can be tested?

Since September 2026, Optimization Experiments have become generally available for Search, Shopping, Audience, and Performance Max. Previously, experimentation capabilities were more limited, primarily by the types of campaigns supported.

Microsoft confirms the ability to test several categories of changes.

Performance Max support is particularly interesting, as automated campaigns are more difficult to evaluate using simple report comparisons.

It's important to distinguish between Optimization Experiments and PMax Uplift Experiments. The former help compare campaign setup variations, while the latter are designed to evaluate the additional conversion impact of Performance Max when used in conjunction with other campaign types.

How to Set Up an Optimization Experiment

Experiments are created directly in Microsoft Advertising.

Path: Campaigns → Experiments → Create → Optimization.

Key setup steps:

Select the source campaign. This will be used as the control variation.

Create an experimental version. Microsoft Advertising creates a separate campaign variation whose settings can be modified.

Definite a testing parameter, such as a bid strategy, targeting, or creatives.

Configure traffic and budget distribution. This is necessary to compare the results of the two variations.

Specify the experiment duration and run the test.

Evaluate results on a single reporting page, where metrics from your control and experimental campaigns are presented side by side.

Microsoft has also simplified the application of results: successful settings can be transferred to the original campaign or used to create a new one.

When planning an experiment, consider the limitations of campaign and budget compatibility. In particular, not all configurations with shared budgets or Dynamic Search Ads are suitable for testing.

Why this is important for advertisers

The main advantage of Optimization Experiments is the ability to make decisions based on controlled tests, rather than changes in metrics between two calendar periods.

For example, after changing the bid strategy, the CPA decreased by 15%. However, this doesn't prove the effectiveness of the new strategy: competition, seasonality, or audience composition could have changed simultaneously.

A/B testing allows you to reduce the influence of such factors and obtain a more informed assessment of the results.

For agencies, this is especially useful when testing new Smart Bidding strategies, optimizing Performance Max, and making major changes to campaign structure.

However, experiments require sufficient budget, duration, and conversion volume. If data is insufficient, the differences between variants may be statistically inconclusive.

Affect's Conclusion

The Optimization Experiments extension is a useful update to Microsoft Advertising, especially for advertisers using automated strategies and Performance Max.

Affect believes that the main value of the tool lies not so much in the ability to compare creatives, but in mitigating risk during significant campaign changes.

We recommend using controlled experiments to test new bidding strategies, switching to automated settings, and other changes that could significantly impact CPA or ROAS.

Before launching, it's important to define the key success metric and ensure that the budget allows for sufficient data to inform decisions.

Sources

Microsoft Advertising — Less busywork, more growth: What's new in Microsoft Advertising this September — https://about.ads.microsoft.com/en/blog/post/september-2026/less-busywork-more-growth-what-new-in-microsoft-advertising-this-september

Microsoft Advertising — AI Max for Search and other product news for August 2026 — https://www.about.ads.microsoft.com/en/blog/post/august-2026/ai-max-for-search-and-other-product-news-for-august-2026