Google expands A/B testing to all Performance Max assets (Beta)

Google has begun rolling out a beta that lets advertisers A/B test asset sets within Performance Max campaigns, giving marketers direct visibility into which creative combinations drive results. The capability — reported by Anu Adegbola at Search Engine Land — “allows advertisers to compare two sets of assets while keeping \u201ccommon assets\u201d consistent across both versions,” a change that could make creative optimization within automated campaigns much more practical.

What this beta does

Performance Max has long centralized ad delivery across Google properties using asset groups and automation, but that very automation made side-by-side testing of creative difficult. The new asset A/B testing feature addresses that gap by letting advertisers create a control arm (assets A) and a treatment arm (assets B) inside the same asset group, while designating any remaining assets as “common assets” that continue to serve to all traffic.

How the test works

According to Google Ads documentation, when you set up an asset comparison test, assets are divided into three categories: the control group, the treatment group and the common assets. Advertisers select which existing or new assets go into each arm and set the traffic split between them. Google then recommends running experiments for a minimum of four to six weeks to reach statistical significance, since Performance Max needs time to exit its learning phase and stabilize delivery.

Key points

• Tests can be created from the Experiments page under the Assets submenu and run inside an existing Performance Max campaign.
• Control and treatment assets count toward asset group limits; locked assets cannot be edited during the experiment.
• New assets submitted for treatment undergo normal policy review and must be approved before they can serve.
• At the end of the test advertisers can apply the treatment assets to the campaign or revert to the control assets — or end the test without changes.

Why this matters

Performance Max’s automation has been a double-edged sword. While it simplifies cross-channel reach and uses Google’s machine learning to optimize delivery, it made it difficult to isolate the impact of individual creative elements. This feature provides a practical lever to test images, headlines and videos without recreating campaigns or relying on shaky workarounds.

For ecommerce and retailers that routinely A/B test product images and promotional creative, asset experiments remove friction and reduce the time between insight and action. As Anu Adegbola notes, the feature “promises a new level of transparency and control over automated campaigns — and could change how marketers approach asset strategy in Performance Max.”

How to approach tests for reliable insights

Advertisers should follow best practices to ensure meaningful results. The Google Ads help center recommends selecting a single asset group per experiment, defining a clear traffic split, and allowing the experiment to run long enough for significance (406 weeks). Locked assets and asset limits mean you should plan the asset set carefully before launching.

Useful tips:

• Start with one hypothesis (for example: new product images vs. existing images) rather than testing multiple variables at once.
• Use similar audience and bidding settings so the experiment isolates creative impact rather than budget or targeting changes.
• Watch for asset approvals — if treatment assets are disapproved your test won’t run as planned.

Implications for advertisers

This beta tightens the feedback loop between creative and performance, letting teams iterate faster. Agencies and in-house teams can embed asset experiments into creative sprints: generate candidate assets, run the experiment, and then rapidly promote winning elements into the live campaign.

However, the need for extended test durations means this feature is best for strategic optimization rather than quick tactical shifts. As web marketer Dario Zannoni observed, “This looks like a very promising feature, and it will be interesting to see how it works and performs in practice,” and he cautioned that tests will likely require considerable time to stabilize.

Actionable next steps

Marketers should take these next steps to evaluate the feature for their accounts:

1. Identify priority asset groups where creative uncertainty exists (top-performing products, new seasonal creative, or messaging variants).
2. Draft a clear hypothesis and select assets for the control and treatment arms; keep unrelated assets as common assets.
3. Schedule experiments for at least 406 weeks and monitor statistical guidance from Google Ads’ Experiment Guidance System.
4. When the experiment ends, apply winning assets and document the change to inform creative playbooks.

Bottom line

Google’s rollout of asset A/B testing for Performance Max fills an important gap for advertisers who want creative-level insights inside automated campaigns. By enabling controlled experiments inside asset groups, the feature combines the reach and efficiency of Performance Max with a methodical way to validate creative choices — provided advertisers plan tests thoughtfully and allow them time to mature.

Original coverage: https://searchengineland.com/google-launches-a-b-testing-for-performance-max-assets-beta-467305