Performance Max A/B Testing: A Playbook for Lead Generation

The short answer

Performance Max A/B testing lets you compare two sets of assets inside one asset group. Google splits traffic between a control arm and a treatment arm, while common assets keep serving to everyone. For lead generation, test one clear message at a time, such as a price-led offer against an expertise-led offer, run it four to six weeks, and judge the winner on qualified leads.

A clean test turns creative opinions into evidence.

Why creative testing pays off for lead generation

In a lead generation account, the asset group is the sales pitch. Headlines, images and videos decide which people raise their hand, and that mix shapes how many of those leads book a visit, tour a school or schedule a service call.

Until recently, Performance Max made that question hard to answer. Asset reports showed performance labels, and teams swapped assets and hoped the change helped. Google now offers a structured way to compare. Its help center describes asset A/B testing for Performance Max as a beta, and in October 2026 Search Engine Land reported a new toggle in the asset group builder that saves your edits as a test.

For a multi-location group, the payoff compounds. A message that lifts booked visits by a few points in one region can roll out across every location.

How Performance Max A/B testing works

Google sorts the assets in the tested asset group into three buckets:

• Control group, assets A: the existing assets you use as the reference.
• Treatment group, assets B: the assets you want to compare, either existing ones or new uploads.
• Common assets: everything else in the asset group, which keeps serving to 100% of traffic alongside both arms.

You set the traffic split between control and treatment. Google's Experiment Guidance System proposes an end date, and Google recommends a minimum of four to six weeks. When the test ends, you choose Apply to move the treatment into the original asset group, or End to discard it.

There are two ways to start:

1. 01Go to Experiments, select Assets, choose Assets provided by you, pick Performance Max and select the campaign and asset group.
2. 02Edit an asset group, scroll to the Save and test asset group changes panel and switch on "Save asset group changes as an A/B test". The button changes from Save to Create test.

On the second path, the option "Apply these changes to the original asset group if the treatment arm wins" is on by default.

7 steps to run Performance Max A/B testing for leads

1. Pick one question worth answering

Strong tests answer a business question. Examples for SalesX's core verticals:

• Dental: same-week appointments against insurance accepted.
• Childcare: safety and curriculum against flexible schedules.
• Pain clinics: provider expertise against fast access to care.
• Local services: upfront pricing against guaranteed arrival windows.

A price-led message against an expertise-led message is a natural first test for most lead generation brands.

2. Change one theme per arm

Keep the treatment focused on a single idea. If arm B carries new headlines, new images and a new video at once, a win tells you little about why. Group assets by message so every asset in arm B tells the same story.

3. Choose a high-volume asset group

Each experiment tests a single asset group. Start with the group that drives the most conversions, so the test reaches a reliable read sooner. A small location group may need the full six weeks or longer.

4. Get the measurement ready first

A test is only as good as its conversion data. Confirm that calls, forms and booking actions are tracked, that only meaningful actions are set as primary goals, and that offline outcomes such as booked visits flow back into Google Ads where possible. Otherwise the test may crown the arm that drives cheap, low-quality form fills.

5. Check campaign eligibility

Google lists situations that block a new experiment: a shared budget, a portfolio bid strategy, Smart Bidding Exploration on the campaign, an overlapping experiment, or an asset group already at the asset limit. Clear these before launch day.

6. Leave the asset group alone during the test

Once the experiment starts, the tested asset group is locked in view-only mode until it ends. Plan seasonal promotions, location openings and policy updates around the test window, and new treatment assets still go through the standard policy review.

7. Decide with the default in mind

With the auto-apply option on, a winning treatment moves into the original asset group when the test ends. That is efficient for clear wins. For brand-sensitive tests, consider switching it off so a strategist reviews the full results, including lead quality from your CRM, before anything goes live.

Which Performance Max experiment fits your question

Google's Performance Max experiments cover more than assets:

• Asset A/B tests compare creative sets within one asset group.
• Optimization experiments test features such as Final URL expansion, and Google notes the tested settings stay on the campaign afterward.
• Uplift experiments measure what Performance Max adds alongside your other campaigns.

Match the experiment to the decision you face. A creative question calls for an asset test, a settings question calls for an optimization experiment, and a budget question calls for an uplift or incrementality study.

How to read Performance Max A/B test results

When the test ends, compare the two arms on the metrics your business runs on:

• Conversions and conversion value for each arm, using only primary actions that reflect real leads.
• Cost per conversion, so a cheaper arm and a better arm are easy to tell apart.
• Lead quality from your CRM: booked visits, tours, enrollments or closed jobs by arm.
• Asset-level performance inside the winning arm, so the next test starts from the strongest pieces.

A treatment that wins on raw form fills and trails on booked visits is a loss in disguise. Join the experiment dates to your CRM before you call a winner. When results are close, treat the test as a tie, keep the stronger brand message and move to the next question.

A quarterly Performance Max testing plan

1. 01Month one: run the strongest message test in the top asset group.
2. 02Month two: apply the winner and roll it to similar asset groups.
3. 03Month three: test a visual theme, such as staff photos against facility photos.
4. 04Each month: log the hypothesis, dates, split, winner and lead quality result in one shared sheet.
5. 05Each quarter: review the log and set the next three questions.

How SalesX turns test results into growth

SalesX runs creative testing as a program across locations, with every test tied to tracked calls, forms and downstream outcomes. That measurement foundation makes each winner trustworthy.

For Children's Learning Adventure, a childcare group with more than 45 locations, the result came from that foundation. In the words of the case study: "Budget scripts across 45+ locations plus call tracking reduced cost per conversion by 49%."

The bottom line

Performance Max A/B testing brings real experiments to the asset group. Ask one business question, change one theme, measure qualified leads and run the test long enough to trust. Then roll each winner across your locations.

Sources and further reading

• Search Engine Land, Google Ads asset group A/B testing toggle spotted in Performance Max
• Google Ads Help, About Performance Max optimization experiments: A/B testing assets (Beta)
• Google Ads Help, About Performance Max experiments

Related reading

• Performance Max for Lead Generation: Consolidate or Segment?
• AI Max for Search: A Testing Playbook for Lead Generation Accounts
• Google Ads incrementality testing
• Children's Learning Adventure case study
• PPC for Multi-Location Businesses: A Scalable Google Ads Framework