Short answer: Google Ads experiments let you test a change on part of a campaign’s traffic and budget before you commit to it. You compare the trial with the original for the dates you set, then apply the change, turn the trial into a new campaign or end it. You find them under Experiments in the Campaigns menu.
How Google Ads experiments work
Google says you can create and run an experiment on a campaign to test the impact of the changes you plan to make. If you split the budget or traffic between the original campaign and the experiment, you can compare how both perform over the dates you set.
If the experiment does better, you can apply it to the original campaign or replace the original. For a custom experiment, Google creates a trial campaign with the new settings, and the trial shares your original campaign’s traffic and budget.
Which experiment fits your change
| Experiment | What it usually tests | Campaigns |
|---|---|---|
| Ad variations | Text ads, responsive search ads or a single change across several campaigns | Search |
| Custom experiments | Smart Bidding, keyword match types, landing pages, audiences and ad groups | Search and Display |
| Video experiments | Which video ad works better on YouTube, with 2 to 4 groups | Video |
Our summary of Google Ads Help, read November 5, 2026.
Google also lists experiments for Performance Max, Demand Gen, AI Max and app assets. Performance Max experiments let you A/B test features, settings and campaigns, and measure the uplift of using Performance Max. If you haven’t used those campaigns yet, start with our Performance Max guide.
Ad variations for your ad text
Ad variations are for Search campaigns. You can find and replace text, for example “Book now” with “Call now”, update headlines and descriptions, or update URLs.
Google splits your ad variation and your original ads by assigning cookies to users, so each person sees only one version, however many times they search. If the ads you test are responsive search ads, our responsive search ads guide explains how they work.
How to set up a custom experiment
- Go to Experiments within the Campaigns menu.
- Select the plus button at the top of the All experiments table and select Custom.
- Choose the original campaigns where you want to test the change.
- Set the change you want to test in the trial campaign.
- Choose the success metrics you want to compare.
- Pick the start date and the duration, and give the experiment a name.
- Finish and schedule it, so it starts on its date.
Google’s setup help describes a 50% split between the original campaign and the trial. A campaign with a shared budget, or one that already has an experiment on the same dates, gives an error: remove the shared budget or pick other dates.
How to read the results
Open the experiment from Experiments and look at the scorecard. Under each metric, the first line shows your experiment’s data. The second line shows the estimated difference with the original campaign, for example +10% for clicks.
It also shows a range for that difference, such as [+8%, +12%], for the confidence interval you choose. The default is 80%, and a blue asterisk marks a statistically significant result.
The Results column names the arm that did better: the control campaign or the treatment campaign. “No clear winner” or “In progress” means the winner can’t be determined yet or there isn’t enough data.
Before you apply the winner
- The experiment ran for 2 to 3 weeks or more.
- The result is statistically significant (it has the blue asterisk).
- You read the range, not only the estimated difference.
- You tested one change, so you know what made the difference.
- Conversion tracking was on, if you compare conversions.
- You chose between applying it and converting it into a new campaign.
RememberAn experiment answers one question with your own traffic. Give it time, read the range next to the estimate, and apply a result only when it carries the blue asterisk.
Apply it, turn it into a campaign or end it
If you apply the experiment, it ends and the changes go into the original campaign. If you convert it, it becomes a new campaign with the same budget and dates, and the original campaign is paused. Google keeps the performance data either way.
Some experiment types can apply favorable results automatically, and that feature is on by default. It then moves 100% of the traffic to the trial campaign, and you can turn it off during the experiment from the Report page.
To stop without applying anything, change the end date so the experiment stops at the close of the day, or select Edit and then End now.
Bid strategies and match types are the usual things to test. Read our Smart Bidding guide before you test a bid strategy, and our guide to keyword match types before you test a match type. Conversion metrics need conversion tracking; our conversion tracking guide shows how to set it up.
If you’d rather hand over the campaign itself, our Google Ads service runs Google Search campaigns, and you keep the ad account in your name. Message us through the form or on WhatsApp.
Quick answers about Google Ads experiments
How long should a Google Ads experiment run?
Google recommends letting an experiment run for 2 to 3 weeks to gather data. If the results still can’t be determined, its Experiments page suggests at least 4 to 6 weeks, and you may need more budget or more time.
What does the blue asterisk mean in an experiment?
It marks a statistically significant result. Google says that means your data is likely not due to chance, and the experiment is more likely to keep performing the same way if it becomes a campaign.
Why isn’t my experiment statistically significant?
Google lists possible reasons: the experiment hasn’t run long enough, the campaign doesn’t get enough traffic, the traffic split was too small, or the changes didn’t make a significant difference in performance.
What happens to my original campaign when I apply an experiment?
If you apply it, the experiment ends and the changes you tested go into the original campaign. If you convert it into a new campaign, the original is paused. Google keeps the performance data either way.
Can I test ad text without a custom experiment?
Yes. Ad variations test changes to Search ads, such as replacing one phrase with another across your ads. Google splits users with cookies, so each person sees only one version of the ad.
Sources
All read on November 5, 2026. Google Ads Help: About the "Experiments" page (formerly drafts and experiments). Set up a custom experiment. Monitor your experiments. Apply a custom experiment. Set up an ad variation.