Google Tests A/B Toggle for Performance Max Asset Groups

Google appears to be testing a built-in A/B testing toggle inside Performance Max asset groups, according to Search Engine Land’s October 20 report.

If you run Performance Max for a high-ticket Shopify store, the catch is the calendar. Google’s help documentation says a test locks the tested asset group and recommends running it at least 4 to 6 weeks, and Black Friday is 37 days out. I cover moves like this every day at Ecommerce Paradise, and this one needs a decision this week, not in November.

Below: what Search Engine Land saw, how asset testing in Performance Max got to this point, what a locked asset group does to a store selling expensive products (if you are new to the model, start with my guide to high-ticket dropshipping), and the dates that decide whether you test at all.

Spending on Google Ads without an LLC behind the store? Form your LLC with Northwest Registered Agent, the best LLC formation service for ecommerce owners. Cheap formation sites hook you with a low first-year price and upsell you at renewal, while Northwest keeps the renewal price the same as year one. Form your LLC with Northwest →

Performance Max Asset Group A/B Toggle: What Google Is Testing

Anu Adegbola, Paid Media News Editor at Search Engine Land, reported on October 20 that Google Ads Coach Joey Bidner spotted a new A/B testing option in the Performance Max asset group interface. Bidner shared the find on LinkedIn.

Search Engine Land wrote that Google “appears to be testing a new asset group A/B testing option,” so availability is not confirmed. The toggle looks built to compare creative approaches inside one asset group. One version might lean on affordability or discounts, and the other on quality or premium service, both promoting the same offer. A meaningful test would need a full set of distinct headlines, descriptions and images for each version, not small tweaks, per the report.

The report leaves gaps. It does not say how Google splits traffic between versions, how results are measured or reported, when or where the toggle rolls out, or who is eligible. It includes no statement from Google.

How the documented asset experiment works

Google’s own help page for Performance Max asset experiments, labeled Beta, describes the route that exists today. You go to Campaigns, then Experiments, press the plus button, choose Assets, then “Assets provided by you,” then Performance Max and “Any assets.”

The test compares two asset sets inside one asset group. Control (A) holds your existing assets, treatment (B) holds the new ones, and common assets serve to 100% of traffic alongside both. Each experiment covers one asset group. You set the traffic split, and the end date comes from Google’s Experiment Guidance System, while the page recommends “a minimum of 4 to 6 weeks.”

Then the lock. Assets in the tested asset group cannot be edited, added or removed until the experiment ends, and the asset pages become view-only. Treatment assets go through standard policy review, and disapproved assets cannot serve. Both arms count toward asset group limits, and the page notes that a group with more than 15 assets can max out.

Eligibility is restrictive. The campaign cannot have a shared budget, sit in a portfolio bidding strategy or use Smart Bidding Exploration, and it cannot have an overlapping experiment on the same dates.

You finish one of two ways. “Apply experiment” adds the treatment assets to the group, with “Keep control assets in campaign” checked by default. “End experiment” discards the treatment assets and reverts the group to its original state. The page does not say which metrics the results report, a gap that matters for high-ticket stores.

Performance Max Asset Testing History: Retail Beta to All Campaigns

This is not the first time the feature has made news. PPC Land reported on January 10 that web marketer Dario Zannoni spotted Google expanding asset-level A/B testing from retail-only Performance Max campaigns, where it launched in October 2024, to all Performance Max campaign types, still in beta. PPC Land quoted Google’s documentation: “When an experiment starts, you can’t edit, add, or remove any assets in the asset group that’s being tested.”

A month later, Search Engine Roundtable reported on February 10 that some advertisers were seeing controlled creative tests in their accounts. Its setup example used Assets A as the live control, Assets B as the treatment and a 50/50 split, run for the typically recommended 4 to 6 weeks.

Then on October 2, Search Engine Land published “Google Ads brings A/B testing to Performance Max asset groups.” It described testing changes to headlines, descriptions, images and videos without first replacing existing creative, and repeated the freeze: “Once an experiment begins, advertisers can’t make additional changes to the asset group while the test is running.”

The three reports do not line up cleanly. One outlet covers in October what two others covered earlier in the year. My read is that this is a staged beta, with accounts getting the feature in waves. Your Experiments tab, not any article date, answers whether you have it.

The upside of a test is simple: the winner is a creative asset you keep running through the first quarter of next year. The risk is the lock, not the idea.

I found no named source disputing the feature’s value. The open questions are Google’s silence on the new toggle and whether it is a shortcut into the same experiments flow or a separate tool. Search Engine Land does not say what the toggle replaces.

Google has been shipping account controls all month. I covered the new channel priority and budget caps last week. The change tying AI Max search terms to headlines and URLs landed the same week.

What a Locked PMax Asset Group Does to Your Black Friday Plan

My read is that this is a creative-testing tool, not a performance lever. It tells you which of two messages your buyers prefer. It does not add budget, fix a product feed or rescue a weak offer.

Here is where it bites. Thanksgiving is November 26, Black Friday is November 27 and Cyber Monday is November 30. An experiment scheduled today starts tomorrow by default, October 22. Four weeks later is November 19. Six weeks later is December 3, which puts the entire holiday peak inside a locked asset group. You could not add a Black Friday headline, swap in the doorbuster image or change the offer text until the test ends. Ending early discards the treatment assets, per Google’s help page.

That makes the useful window narrow. A four-week test has to start by October 28 to finish on November 25, the day before Thanksgiving. Start later and you either test through the sale or cut the test short, and a cut-short test teaches you nothing. My Black Friday Google Ads setup guide covers what has to be live before the 25th anyway.

Now the volume problem. This is hypothetical math, not reported data. Say a store spends $300 a day on one asset group for 28 days, which is $8,400. At a 5x return that is about $42,000 in revenue, and at a $2,000 average order that is 21 orders. A 50/50 split puts roughly 10 or 11 orders in each arm. Counts that small swing by about the square root of themselves, so one arm showing 8 orders and the other 13 can easily be chance.

If Google’s results report purchases only, a high-ticket store should expect noise. Tracking is also softer than it was, as my Safari 27 audit showed, which makes small gaps even less trustworthy.

Here is my rule of thumb, not Google’s. With 40 or more orders a month in the asset group, a four-week test can show a real gap if the creative contrast is big. Between 15 and 40, test only a dramatic contrast, such as price-led against white-glove service. That is the same tension I wrote about in how to scale a high-ticket store without cutting prices. Under 15 orders, skip the experiment and make one deliberate change at a time, with notes.

Check eligibility before you fall in love with the idea. A shared budget, a portfolio bidding strategy or Smart Bidding Exploration blocks the experiment, and a group near the 15-asset ceiling may need trimming first, per PPC Land’s summary of the documentation. A Shopping feed that changes prices every few days has its own risk, because Google may disapprove products with fast-changing prices.

Building two truly different asset sets is the real work. What I’d do is draft the headline and description variants with Claude.

Then build the image variants in Canva, with one set showing the product at its sharpest price and the other showing delivery, warranty and white-glove setup. Before you write the premium version, read what your rivals say in their own ads. Semrush shows competitor ad copy, which keeps you from testing a message everyone in your category already runs.

Since Google’s page does not say which metrics the results report, build your own scoreboard. Record Shopify orders, revenue and gross margin by day for the full test window, and note which creative was live. If the toggle turns out to be a shorter path into the same experiment, the lock and the 4 to 6 weeks probably come with it, but that is an assumption until Google documents it.

A premium-positioned winner also changes your inbox. If the white-glove version wins on a big-ticket item, expect more pre-sale questions, so staff chat with Gorgias before the traffic arrives. For delivery updates once orders flow, AfterShip keeps buyers off your phone line.

Demand is already here. Shoppers are starting in October, per the numbers in my Deloitte holiday shopping breakdown, so the traffic feeding either arm is arriving now.

Whichever arm wins, the shoppers who click and leave still need a second touch. A Klaviyo browse-abandonment flow handles the email side. Phone and form leads need a plan too, which is what my guide to high-ticket sales follow-up covers.

If building two full creative sets, running the experiment and staffing the sale sounds like a second job, that is exactly what my team’s done-for-you turnkey service exists to take off your plate.

Four weeks of locked ads, a sale 37 days out and no room to be wrong. Want my team to scale the store you already have? See the scaling service →

PMax Asset Test Checklist: Launch by Oct. 28 or Wait Until January

Five moves, in this order:

Pull your order count per asset group for the last 60 days and apply the thresholds above. Under 15 orders a month, skip the test and spend the energy on the sale instead.
Open Campaigns, then Experiments, in Google Ads and confirm the campaign qualifies. Remove any shared budget, portfolio bidding strategy or Smart Bidding Exploration, and trim any asset group sitting near 15 assets.
Pick your highest-order asset group and write two clearly different versions, such as price-led against premium service, each with its own headlines, descriptions and images. Run the premium version past your Shopify product pages so the landing experience matches the promise.
Schedule the test to start no later than October 28 so it ends November 25. Build your Black Friday creative in a separate asset group that is not under test. Nothing in the documentation I read stops that, but confirm it in your own account, and run through my high-ticket Black Friday checklist while you are in there.
Log the start and end dates next to every other account change, and track margin with Finaloop so a creative win that quietly costs you profit does not slip by. If you want a second set of eyes on the account before the 28th, book a call on my discovery page.

Frequently Asked Questions

Is the asset group A/B toggle live for everyone?
No. Search Engine Land says Google appears to be testing it, with no rollout details and no Google statement. The Experiments route in Google’s help documentation is the documented way to test assets today.

How long should a Performance Max asset test run?
Google’s help page recommends a minimum of 4 to 6 weeks. Starting October 28 gives you four weeks and finishes before Thanksgiving.

Can I edit assets while the test runs?
No. The tested asset group is view-only until the experiment ends. Ending early discards the treatment assets and reverts the group.

Which campaigns cannot run an asset experiment?
Per Google’s help page, campaigns with a shared budget, a portfolio bidding strategy or Smart Bidding Exploration. Remove those first, or test a different campaign.

Does asset testing work for a store with low order volume?
It gets noisy. With roughly 10 orders per arm, chance can look like a winner. Fix the conversion path first, starting with my post on why high-ticket Google ads need a complete conversion system.

Should I send high-ticket traffic to Performance Max at all?
That depends on your feed quality and margins. I laid out the Shopping side in my guide to turning Google Shopping clicks into high-ticket sales. My free beginner guide covers the basics.

Does the winning arm apply automatically?
Not according to Google’s help page. You choose “Apply experiment” to add the treatment assets, with control assets kept by default, or “End experiment” to discard the test and revert.

Testing ads is easier with someone who has run these accounts before. Want 1-on-1 coaching to launch or fix your high-ticket store? Get the coaching details →

Run the numbers on your own asset groups before you touch the Experiments tab, and decide before the 28th. Subscribe to the YouTube channel for daily breakdowns. More breaking news coming through the day. I wish you guys the best of luck out there.

If this was useful, these go deeper:

Trevor Fenner

Trevor Fenner is an ecommerce entrepreneur and the founder of Ecommerce Paradise, a platform focused on helping entrepreneurs build and scale profitable high-ticket ecommerce and dropshipping businesses. With over a decade of hands-on experience, Trevor specializes in high-ticket dropshipping strategy, niche and product selection, supplier recruiting and onboarding, Google & Bing Shopping ads, ecommerce SEO, and systems-driven automation and scaling. Through Ecommerce Paradise, he provides free education via in-depth guides like How to Start High-Ticket Dropshipping, advanced training through the High-Ticket Dropshipping Masterclass, and fully done-for-you turnkey ecommerce services for entrepreneurs who want a faster, more hands-off path to growth. Trevor is known for emphasizing sustainable, real-world ecommerce models over hype-driven tactics, helping store owners build scalable, sellable, and location-independent brands.