How to A/B Test AdSense Ads for Higher Earnings (AdSense Experiments)

A/B testing AdSense ads is the surest way to find out which setup earns most on your site, instead of guessing. In 2011 we showed how to split traffic between two ad units with different link colours, using a little JavaScript to show each one half the time and custom channels to compare results. That trick is outdated: colour options are gone, and altering ad code is risky. Today AdSense has a built-in Experiments tool that does the split testing for you. Here is how to use it in 2026.

Quick answer: In AdSense, open Optimization > Experiments, create an experiment (for example Auto ads settings, ad balance or blocking controls), choose the site and let Google split traffic between your original and the variation. Wait until AdSense reports a clear result, then apply the winner. Test one change at a time, and judge by page RPM, not just CTR.

What you can test with AdSense Experiments

• Auto ads settings: ad load, formats such as anchors, side rails and vignettes, and excluded areas.
• Ad balance: showing fewer, better-paying ads can keep earnings while improving user experience.
• Blocking controls: see whether unblocking certain categories or advertisers raises revenue.

Read our earlier guide to increase AdSense CTR with AdSense Experiments.

How to run an AdSense experiment

1. Sign in to Google AdSense and go to Optimization > Experiments.
2. Click to create a new experiment and pick the type.
3. Select your site and set the variation’s settings.
4. Choose whether to apply the winner automatically, then start the experiment.
5. Let it run until AdSense shows a confident result, often a couple of weeks depending on traffic.

Testing manual ad placements

Experiments do not cover every change, such as moving a manual ad unit higher on the page. For those:

• Create a separate ad unit (or custom channel) for each placement so you can compare them in reports.
• Change one placement at a time and compare equal periods, such as two full weeks, to smooth out weekday swings.
• Compare page RPM and viewability, since AdSense now pays mainly per impression.
• WordPress ad plugins like Advanced Ads offer built-in A/B rotation that keeps ad code unchanged.

Avoid hand-written JavaScript that rewrites AdSense code, as our 2011 example did; changing the ad code can break AdSense policies. Stick to official tools and approved plugins, and avoid these common AdSense mistakes.

What to measure

Ideas worth testing

• Fewer ads versus more ads on long articles.
• Anchor ads on and off for mobile.
• An in-article ad after the first paragraph versus after the third.
• Responsive units versus fixed sizes in the sidebar.

For placement ideas, see our guide to AdSense heat map and best placements, and how to increase AdSense CTR.

Common A/B testing mistakes

• Changing several things at once: you will not know which change made the difference.
• Stopping too early: a few days of data is usually noise, especially on smaller sites.
• Testing during unusual weeks: festivals, sales seasons and viral traffic spikes distort results.
• Ignoring readers: a variation that earns slightly more but makes pages cluttered can cost you returning visitors over time.

FAQ

Is A/B testing allowed in AdSense?

Yes, through AdSense Experiments or approved ad management tools that do not alter the ad code.

How long should an AdSense experiment run?

Until AdSense shows a clear result, usually one to four weeks depending on traffic.

Can I still test link colours?

No. AdSense now styles ads automatically, so colour tests are no longer possible.