Microsoft Introduces Performance Max Experiments: A New Era of Testing for Automated Campaigns

In a significant move designed to provide digital marketers with greater control and transparency over their automated ad spend, Microsoft Advertising has officially expanded its experimentation platform to support Performance Max (PMax) campaigns. This update represents a major shift for advertisers utilizing Microsoft’s network, offering them structured, risk-free methodologies to test campaign optimizations, measure incremental growth, and evaluate the performance of automated formats without risking active, live campaign budgets.

Previously, experimentation within the Microsoft Advertising ecosystem was restricted almost exclusively to traditional Search campaigns. By extending these capabilities to Performance Max, Microsoft is addressing a critical pain point for modern advertisers: the "black box" nature of AI-driven, multi-channel campaigns.

1. Main Facts: The New PMax Experimentation Framework

Microsoft Advertising’s latest update introduces dedicated testing capabilities directly into the campaign management dashboard. Advertisers can now set up controlled experiments to isolate variables and compare different setups within their Performance Max campaigns.

The Core Features of the Update

• Availability: The new testing capabilities are located within the Microsoft Advertising platform under the Campaigns > Experiments tab for all eligible accounts.
• Experiment Types: Microsoft has introduced two primary experiment structures designed specifically for Performance Max. These allow advertisers to either test a Performance Max campaign against standard campaign types (such as Standard Shopping or Search) to measure incrementality, or to test different asset groups, bidding strategies, and targeting parameters within a PMax environment.
• Platform Rebranding: To accommodate this expansion, Microsoft has rebranded its legacy testing suite. The existing experimentation features have been renamed Search optimization experiments, clearly distinguishing them from the newly introduced Performance Max experiments.
• Discovery: The update was first identified in Microsoft’s official technical documentation by Hana Kobzová, a prominent pay-per-click (PPC) expert and the founder of PPC News Feed.

By providing a native testing environment, Microsoft allows advertisers to split their target audience or search traffic into mutually exclusive groups. This split-testing framework ensures that there is no overlap between the control group and the experiment group, yielding clean, statistically significant data.

2. Chronology: The Evolution of Automation and Testing on Microsoft Ads

To understand the importance of this update, it is essential to trace how Microsoft’s advertising ecosystem has evolved from a manual, search-centric network into an AI-powered, multi-channel platform.

[2020-2022: Search-Centric Testing] 
Microsoft Ads limits experiments to standard Search campaigns. 
       │
       ▼
[2023: The Launch of Microsoft PMax] 
Microsoft rolls out Performance Max globally to compete with Google's automated offerings.
       │
       ▼
[Early 2024 - 2025: The "Black Box" Challenge] 
Advertisers demand more transparency and control over automated budget allocation.
       │
       ▼
[Mid-2026: The Launch of PMax Experiments] 
Microsoft rebrands legacy testing to "Search optimization experiments" and rolls out PMax Experiments.

The Era of Manual Control and Search-Only Testing

For over a decade, digital advertising relied heavily on manual keyword bidding, precise match types, and granular ad group structures. During this period, Microsoft Advertising (formerly Bing Ads) provided robust A/B testing tools, but these were strictly confined to Search campaigns. Advertisers could easily split-test landing pages, ad copy, and bidding strategies because the variables were highly structured and predictable.

The Rise of Performance Max

With the rapid advancements in machine learning, search engines pivoted toward automated, cross-channel campaign formats. Google launched its version of Performance Max first, with Microsoft following suit to remain competitive.

Microsoft’s Performance Max campaigns were designed to streamline ad creation and distribution across the entire Microsoft network—including Bing Search, the Microsoft Start feed, MSN, Xbox, Outlook, and the Microsoft Audience Network (MSAN). PMax utilized artificial intelligence to dynamically mix and match headlines, descriptions, images, and videos, bidding in real-time to find high-value converting customers.

The Transparency Gap

While PMax campaigns yielded impressive scale, they introduced a major challenge: a lack of transparency. Advertisers struggled to determine whether PMax was driving truly incremental conversions or simply cannibalizing traffic from their existing, highly optimized Search and Shopping campaigns. Without native testing tools, running an A/B test on PMax required manual, high-risk workarounds, such as pausing campaigns sequentially or setting up complex geographic split tests, both of which frequently disrupted machine learning algorithms and degraded performance.

The launch of Performance Max experiments in mid-2026 represents the resolution of this multi-year challenge, bridging the gap between automated scaling and statistical validation.

3. Supporting Data: Why Testing in Performance Max is Critical

Automated advertising campaigns rely heavily on historical data and continuous learning. When an advertiser makes a sudden, major change to a live PMax campaign—such as adjusting target return on ad spend (tROAS), changing target cost per acquisition (tCPA), or swapping out creative assets—the campaign can easily enter a destabilizing "re-learning" phase. During this period, performance often drops, resulting in lost revenue and wasted ad spend.

The Risk of Manual Changes vs. Controlled Experiments

Industry benchmarks highlight why controlled experimentation is financially and operationally superior to making direct modifications to live campaigns:

The Power of Split Testing

By utilizing Microsoft’s native split-testing infrastructure, advertisers can allocate a specific percentage of their traffic (e.g., 50%) to the experimental PMax setup while leaving the remaining portion on the control setup.

This parallel execution ensures that external market factors—such as a sudden holiday shopping surge, competitor bid adjustments, or macroeconomic changes—impact both the control and experiment groups equally. Consequently, any observed difference in conversion rate, cost per conversion, or return on ad spend can be confidently attributed to the changes being tested.

4. Expert Viewpoints and Technical Setup

The introduction of this feature has been met with enthusiasm across the digital marketing community, particularly among enterprise-level agencies and performance marketers who manage large, multi-channel budgets.

Hana Kobzová’s Discovery and the PPC Community’s Reaction

Hana Kobzová, founder of PPC News Feed, was the first to flag the updated help documentation detailing the feature. Her discovery quickly sparked discussions among search marketing professionals.

Historically, advertisers have been wary of fully automated campaigns because they operate as "black boxes." The consensus among PPC specialists is that while automation is highly effective at scale, it requires guardrails.

Anu Adegbola, Paid Media Editor of Search Engine Land and a veteran PPC strategist, has frequently emphasized the importance of rigorous testing in automated environments:

"In an era where machine learning dictates ad placements and bidding, advertisers cannot afford to fly blind. The ability to test and validate the incremental lift of automated campaigns like Performance Max is no longer a luxury—it is a fundamental requirement for responsible budget management."

How to Access and Set Up PMax Experiments

According to Microsoft’s updated help documentation, setting up a Performance Max experiment involves a straightforward, wizard-guided workflow:

1. Navigate to Experiments: Log into the Microsoft Advertising online portal and navigate to the left-hand navigation menu. Click on Campaigns, and select Experiments.
2. Create a New Experiment: Click on the plus icon to create a new test and select Performance Max experiment as the experiment type.
3. Define the Control and Treatment: Select your existing "base" Performance Max campaign as the control. Then, define the treatment parameters—such as testing a new asset group, modifying audience signals, or adjusting target ROAS.
4. Allocate Traffic and Budget: Choose the split ratio (e.g., 50% control, 50% experiment) and define the duration of the test. Industry standards suggest running experiments for at least 4 to 6 weeks to gather sufficient data and account for conversion delays.
5. Monitor Results: Microsoft’s reporting dashboard will display real-time performance comparisons, complete with statistical significance indicators to help advertisers determine when a test has yielded a clear winner.

5. Strategic Implications for the Digital Advertising Landscape

The rollout of Performance Max experiments has broad strategic implications for advertisers, digital agencies, and the competitive dynamic between Microsoft and its primary rival, Google Ads.

1. Achieving Feature Parity with Google Ads

For years, Microsoft Advertising has positioned itself as an essential secondary network for search marketers, offering access to high-value audiences that are often cheaper to reach than those on Google. However, to capture a larger share of global ad spend, Microsoft must consistently offer the same level of sophisticated tooling as Google.

Google Ads has supported PMax experiments for some time, allowing advertisers to run incrementality tests against Standard Shopping campaigns. By closing this feature gap, Microsoft removes a major barrier to entry, making it significantly easier for enterprise brands to mirror their Google testing frameworks directly within the Microsoft ecosystem.

2. Empowering Agencies to Prove Incremental Value

For digital marketing agencies, proving the value of their services to clients is a constant challenge, particularly when using highly automated campaigns. Clients often worry that automated campaigns are simply bidding on brand keywords or taking credit for organic conversions.

With PMax experiments, agencies can set up clean, data-backed incrementality tests. For example, an agency can run an experiment testing a PMax campaign against standard search campaigns to prove exactly how many additional conversions the automated format is driving. This data-driven proof is crucial for justifying marketing budgets and building long-term client trust.

3. Mitigating the Risks of Creative Fatigue and Asset Optimization

Performance Max campaigns rely heavily on creative assets (images, videos, headlines, and descriptions) to generate high-performing ad combinations. Over time, these assets can suffer from "creative fatigue," leading to a decline in conversion rates.

With the new experimentation tools, advertisers can systematically test new creative asset groups against older ones in a controlled split test. This allows brands to refresh their visual identity and messaging continuously, scale successful creatives, and retire underperforming ones without risking a sudden drop in overall account performance.

4. Accelerating the Transition to AI-Driven Search

Ultimately, Microsoft’s update is designed to make advertisers more comfortable with automation. By providing a safety net in the form of robust experimentation, Microsoft is encouraging traditional search marketers to transition their manual, keyword-based campaigns over to AI-driven formats. As advertisers gain confidence in their ability to test and control PMax, adoption rates are expected to rise, further cementing automated, multi-channel campaigns as the standard for digital advertising.