Performance Max asset groups can now be A/B tested, which means you can finally prove whether a creative change helped instead of guessing. The feature was spotted in accounts on October 2, 2026. There is one restriction that will change how you plan your tests, so read that part before you start one.
What You Can Test Now
The experiment runs inside the asset group, which is where PMax creative actually lives. You can test changes to headlines, descriptions, images, and videos against what you are running today.
That covers most of the creative questions that come up in a Performance Max campaign, and it is the first time Google Ads has given you a controlled way to answer them inside PMax.
- New creative concepts against your existing assets. The straightforward version of the test.
- UGC-style content against polished brand creative. This one has been an argument in every account I have worked in, and now it has an answer.
- Different messaging or image styles. Swap the angle, not just the wording.
- The impact of adding or changing video. Worth running on its own, because video is the asset most advertisers skip.
The Restriction That Matters
Once an experiment starts, you cannot make any other changes to that asset group while the test is running. Google locks it so other edits do not contaminate the results.
That is the right call statistically, and it is going to frustrate people who optimize weekly. You cannot start a test on Monday and then swap a headline on Thursday because something looked off. You are committing the asset group for the length of the experiment.
So plan the test before you launch it. Decide what you are changing, decide how long it runs, and get everything else in that asset group the way you want it first. If there is a fix you have been meaning to make, make it before you hit start, not after.
Where This Came From
This one was not a formal Google announcement. It was spotted in live accounts by paid search specialist Bia Camargo, who shared it on LinkedIn, and picked up from there.
That is worth knowing for two reasons. Features that surface this way are often still rolling out, so if you do not see it in your account yet, that is normal. And the details can shift before Google documents them formally.
It fits a pattern though. Google has been steadily adding reporting and controls to Performance Max as advertisers push for more transparency, and this is the most useful one in a while.
My Take
I have been skeptical of PMax creative advice for a long time, and the reason is simple. Nobody could prove any of it. You would swap your images, performance would move, and you had no way to separate your change from seasonality, from auction pressure, or from Google shifting budget between channels on its own.
A controlled test fixes that. Not completely, but enough that you can stop arguing from opinion.
One of the biggest mistakes I see with A/B testing in Google Ads is calling a result too early. Give the experiment enough conversion volume to say something real. A creative test that ran for six days on eleven conversions has not told you anything, and acting on it is worse than not testing at all.
How I Would Run The First One
- Pick one variable. Images or headlines or video. Not all three, or you will not know what moved.
- Clean up the asset group first. You are locked out once the test starts, so fix anything else now.
- Test the thing you actually disagree about. If the debate in your account is whether UGC beats studio creative, test that. Do not burn the first experiment on something nobody cares about.
- Set the end date before you start. Decide what volume you need to call it, and hold yourself to it.
- Write down what you expect to happen. It keeps you honest when the result is boring, which it will be some of the time.
Keep in mind, a creative test tells you about creative. It will not fix a feed problem, a conversion tracking problem, or a budget that is too small for the campaign to learn anything in the first place.
Question to Answer:
What is the one creative assumption in your Performance Max account that you have never actually tested, and do you have enough conversion volume to test it properly?
In Summary
Performance Max asset groups now support A/B testing for headlines, descriptions, images, and video. The catch is that the asset group is locked to other edits while the experiment runs, so plan the whole thing before you launch it.
Start with the creative question you argue about most, give it enough volume to mean something, and resist the urge to call it in week one.
0 comments