Google Ads may be rolling out a toggle called “Save and test asset group changes,” which turns an edit to headlines, descriptions, images or videos into an A/B test. The report comes from a screenshot that advertiser Joey Bidner posted on LinkedIn, covered by Search Engine Roundtable on October 21, 2026. Google has not confirmed a launch.
What showed up on screen?
According to Search Engine Roundtable, the screenshot shows a switch inside an asset group, the bundle of headlines, descriptions, images and videos that Performance Max mixes into ads. The setting says it measures the performance impact of changes to those assets, and it warns that you will not be able to change the asset group while the experiment is running.
Bidner said the idea looks useful for testing two messaging angles for the same product or service, with the catch that you need a full set of alternative creative: different headlines, images and descriptions.
What is still unknown?
The story rests on one screenshot and no official documentation. Nobody has confirmed which accounts get the option, when it will launch, or how traffic would be split. Treat it as a feature in testing that could change names or never reach every account.
Does asset testing already exist in Performance Max?
Yes. Google’s help page describes an A/B test for Performance Max assets, currently in beta. It compares two sets of assets inside one asset group: the control (A), which is what you run today, and the treatment (B), made of the assets you pick or upload. Anything left out of both sets is a common asset and keeps serving to all traffic.
PPC Land reported that the test launched in October 2024 for retail campaigns only and was expanded in January 2026 to every Performance Max campaign type, including lead generation and local sales.
How do you set up the test today?
The path in Google’s documentation looks like this:
- In the Campaigns menu, open Experiments and click the plus button.
- Choose Assets, then “Assets provided by you.”
- Select Performance Max as the campaign type.
- Pick the campaign and the asset group to test.
- Choose the treatment assets. You can add to the existing ones or replace them.
- Set the traffic split between control and treatment, name the experiment and schedule it.
The start date defaults to the next day. Google recommends running the test for 4 to 6 weeks, and the end date is suggested by its Experiment Guidance System.
What are the limits?
- Each experiment tests one asset group.
- While the test runs, the group is locked: you cannot edit, add or remove assets until it ends.
- Control and treatment assets both count toward the group’s limits. If the group has more than 15 assets, you may need to remove some first.
- New assets go through policy review. A disapproved asset cannot serve in the experiment, which can skew the result.
- The campaign cannot use a shared budget, sit in a portfolio bidding strategy or use Smart Bidding Exploration.
When the test ends, you either apply it, which moves the treatment assets into the original group (with the option to keep or drop the control assets), or end it without changing anything.
What does this change for small advertisers?
This part is our opinion, not something the sources say. Structured testing is good, but a test only means something when each side collects enough conversions. A made-up example to show the math: if a campaign produces 6 sales a week and you split traffic 50/50, each side gets about 3 sales a week, or 18 over six weeks. A gap of 2 or 3 sales between the sides fits inside normal luck, so it cannot settle a decision.
If your conversion volume is low, test one big change at a time (a different offer, not a swapped comma), let the test run for the full recommended period and avoid touching bids or budget meanwhile. If you already judge campaigns by break-even ROAS, reading the result gets easier: side B only wins if it lifts sales without hurting your margin.
It also helps to check your spend before you start. With a small budget, see how to split it between Google and Meta so you do not spread a little money across too many experiments.
What should you do now?
Open the Experiments page in your account and check whether asset testing appears. If it does, pick the asset group with the most conversions, build a complete alternative set (headlines, descriptions and images with a different angle) and write down what counts as a win before you begin, for example a lower cost per sale. If the “Save and test” toggle shows up inside an asset group, expect it to follow the same logic as the experiment Google documents, but confirm the rules on screen, since the feature has no official documentation yet.
