Pinterest Launches Self-Serve A/B Testing: How to Create and Run Experiments
Pinterest introduces Self-Serve A/B Testing: Advertisers can conduct their own experiments in Ads Manager.
In September 2026, Pinterest expanded its A/B testing capabilities for advertising campaigns, making experiments available for self-service setup directly in Ads Manager. The new Self-Serve A/B Testing tool allows advertisers to compare creatives, audiences, optimization strategies, and Pinterest Performance+ campaigns using controlled audience splits and statistical evaluation of results.
Pinterest had previously tested this tool in a limited mode. Specifically, in Q1 2026, the company announced the launch of built-in A/B Testing in beta for comparing Performance+ with traditional campaigns. By September, training materials and detailed documentation on setting up experiments were available.
What is Pinterest Self-Serve A/B Testing?
Self-serve A/B Testing is a built-in tool for experimentally comparing Pinterest ad campaigns. Its purpose is to determine whether a change in advertising strategy actually improves performance, rather than simply resulting in random fluctuations.
Standard advertising reporting shows how many clicks, purchases, or revenue a campaign generated. However, comparing two campaigns solely by CPA or ROAS doesn't always allow for objective conclusions. Results may vary due to audience composition, competition intensity, timing, and other factors.
A/B testing solves this problem through a controlled experiment.
The advertiser creates two or more campaign variations that differ in a specific parameter. Pinterest randomly assigns users to test groups so that each user participates in only one variation of the experiment. The system then compares the results across the selected metrics and evaluates the statistical significance of the differences.
For example, an online store might test which strategy delivers the best ROAS:
Version A: a standard Shopping campaign with manual settings.
Version B: a Pinterest Performance+ campaign with automated optimization.
Both campaigns use comparable budgets, dates, product offerings, and measurement conditions. Pinterest splits the audience between the variants and generates a report upon test completion.
Therefore, Self-Serve A/B Testing is not simply a tool for comparing two performance graphs, but a system for conducting controlled advertising experiments.
What variables can be tested with A/B testing on Pinterest?
Pinterest allows you to choose one of four main experiment types. Each is designed to test a specific set of hypotheses.
Creative — images, videos, ad formats, landing pages, Performance+ Creative settings.
Targeting — audiences, demographics, placements, and Performance+ Targeting options.
Objective, Budget, or Optimization — campaign goals, budget settings, bidding strategies, and optimization.
Pinterest Performance+ Campaign — automated Performance+ Campaign versus a standard campaign.
Creative Testing
An advertiser can test which visual solutions drive higher engagement or conversions.
For example, a clothing store might compare two approaches to promoting the same collection:
Version A: static product images.
Version B: short videos showcasing clothing.
Targeting, budgets, and other settings remain consistent.
It's important to note that Pinterest recommends testing significantly different concepts. Changing a small headline element or background color may have too little effect to confidently determine a winner.
Targeting Testing
This experiment evaluates different approaches to audience discovery.
For example:
Version A: user interests related to home decor.
Version B: a broader audience with automated targeting.
The goal of testing is to determine which approach results in higher-quality traffic or better conversion rates.
Objective, Budget, or Optimization Testing
This type of testing is used to compare advertising campaign optimization methods.
For example, an advertiser can evaluate how performance changes when using different objectives or bid combinations.
However, Pinterest recommends not mixing fundamentally different changes in a single experiment. Changing audiences, creatives, and optimization strategies simultaneously can make it difficult to determine the cause of changes in results.
Pinterest Performance+ Campaign Testing
One of the most interesting scenarios is comparing a standard campaign with Pinterest Performance+.
Pinterest Performance+ is a set of automated tools that use algorithms to optimize targeting, bidding, and individual elements of an advertising campaign.
When conducting an A/B test, an advertiser can create a standard campaign and its Performance+ alternative with comparable conditions.
This experiment allows you to verify whether automation actually improves the results of a specific advertising account, rather than relying solely on general platform claims.
For this type of test, Pinterest recommends maintaining the same creatives, goals, budgets, and dates to isolate the impact of the Performance+ strategy itself.
Where and how to set up A/B Testing in Pinterest Ads Manager
The new tool is located directly in your Pinterest ads account. Creating an experiment doesn't require a separate analytics platform or contacting a Pinterest representative.
Setup is done through the Experiments section.
Path: Business Account → Analyze Performance → Experiments → Create A/B Test.
Step-by-step experiment setup
Create an experiment. Open Experiments and click Create A/B Test. Enter a name and select the Test Variable—the parameter you want to test.
Select performance metrics. In the Test Goal section, set the Primary Metric and Secondary Metric. For example, the primary metric is ROAS, and the secondary metric is CTR. The winner is determined based on experiment data, so the metric you choose should align with your business goal.
Set dates. In the Test Schedule, specify the start and end dates for the experiment. Create and publish it at least 24 hours before the start.
Create versions. Pinterest allows you to use existing campaigns (Use Existing Campaign) or create variants by copying them (Copy a Campaign). In the latter case, select the control Version A, then create Version B with the modified parameter.
Review the variant settings. All essential parameters, except the one being tested, must be identical. If necessary, add Version C and Version D. You can compare up to four versions.
Check the Estimated Test Power. Pinterest evaluates whether the budget and expected data are sufficient to obtain a statistically valid result. If the test status is Weak, it is recommended to adjust the budget and update the forecast. The preferred status is Strong.
Run the test. After verification, click Publish. Further audience and results comparisons are performed using Pinterest tools.
Pinterest allows you to create variants from campaigns with multiple ad groups, but recommends using one Ad Group per campaign for a cleaner experiment.
What limitations should be considered?
Duration and Budget
Pinterest recommends running tests for 14 days to two months. Metrics related to purchases and revenue may require longer than comparing CTR or CPC.
The budget is calculated for each version for the entire testing period. The recommended spend depends on the market, the selected metric, and the type of experiment.
For example, the official documentation for North, Central, and South America provides the following minimum budgets for one version:
Engagement Metrics — $825–$5,559.
Conversion Metrics — $825–$18,000.
This is not a fixed A/B Testing cost or a mandatory minimum threshold for all advertisers. The system calculates recommendations for a specific experiment. For two variants, the total budget will be the sum of the costs of both campaigns.
Structure Limitations
All variants must operate in the same market and have comparable conditions. It is recommended to change one main group of parameters within an experiment. When testing different creatives, audiences, and optimization goals simultaneously, the results may be statistically significant, but the advertiser will not be able to determine which specific factor led to the improvement.
It's also important to understand that when there are multiple creatives within a single version, Pinterest evaluates the results of the version as a whole, not each ad individually.
How Pinterest Determines a Winner
After a test is complete, Pinterest processes the data. Typically, a report appears within approximately 48 hours.
You can view it in the Experiments → Manage Tests → View section.
The final report displays the values of the primary and secondary metrics, results graphs, and the Winning Probability.
Pinterest may declare a winner if the probability of one version being superior exceeds 85%. If there is insufficient data or small differences, the experiment may end without a clear winner.
For example, if Version B achieved a ROAS of 4.2 versus Version A's 3.8, this does not necessarily mean it will be recognized as the best. To draw a confident conclusion, it is necessary to consider the statistical evaluation of the differences.
If a winner is undecided, advertisers are advised to review sample size, duration, budget, and differences between options, rather than automatically assuming both strategies are equally effective.
Who it applies to and why it's important
Self-serve A/B Testing is especially useful for ecommerce advertisers, DTC brands, and agencies that regularly test different creative concepts and buying strategies.
For brands with a large product range, Pinterest is particularly valuable as a visual platform where ad creative, image style, and product presentation can significantly influence user behavior.
The tool also allows for systematic evaluation of the transition to Performance+. Instead of comparing reports for two different periods, advertisers can test an automated strategy against an existing configuration in a controlled experiment.
For advertising agencies, this means a more transparent way to justify changes to an advertising strategy based on testing results.
A/B Testing does not guarantee increased effectiveness. Its advantage is that it reduces uncertainty when choosing between alternative settings.
Affect's Conclusion
The launch of Self-Serve A/B Testing is an important development in Pinterest Ads' measurement tools. The platform gives advertisers the ability to independently conduct controlled experiments that previously required significantly more manual work or specialized research.
Affect believes that the greatest value lies not in the ability to compare two ad images, but in the comprehensive testing of optimization strategies and Pinterest Performance+.
Specifically, advertisers considering switching from traditional campaigns to Performance+ now have a more robust way to evaluate the impact of automation on CPA and ROAS.
However, we recommend considering the statistical power of an experiment early in the planning stage. For smaller ad accounts, budget and event requirements can be a significant constraint.
Affect recommends starting with a clearly formulated hypothesis, choosing one primary business metric, maintaining consistent settings across variations, and making decisions only after completing the test and analyzing the statistical significance of the results.
Sources
Pinterest Business Help — A/B Testing: https://help.pinterest.com/en/business/article/ab-testing
Pinterest Business Help — Create and Manage A/B Tests: https://help.pinterest.com/en/business/article/create-and-manage-ab-tests
Pinterest Business Help — Review A/B Test Results: https://help.pinterest.com/en/business/article/review-ab-test-results
Pinterest Academy — Run A/B Testing on Pinterest: https://www.pinterestacademy.com/student/path/3639177-run-a-b-testing-on-pinterest
Pinterest Q1 2026 Earnings Transcript: https://s204.q4cdn.com/369458543/files/doc_earnings/2026/q1/transcript/Q1-2026-Transcript.pdf
