You don’t necessarily have to launch a new ad format across 100% of your traffic right away. It’s best to test it first.

But an effective test isn’t just about enabling the new format and looking at the eCPM. It’s important to define the experiment’s conditions, metrics, and success criteria in advance.

Step 1. Start with a small portion of traffic

For the first test, you can allocate 10-20% of your audience, leaving the remaining traffic as a control group.

For example, one group of users sees the familiar ad format, while the other sees the new version. All other conditions should remain as consistent as possible.

This approach allows you to test your hypothesis without immediately changing the advertising strategy for the entire site.

Step 2. Define Key Metrics

During the first 72 hours, it’s important to look beyond just ad revenue.

Viewability shows how often an ad unit actually appears within the visible area of the screen.

Time-on-Site helps assess whether the time users spend interacting with the site has changed.

Bounce Rate shows whether the percentage of visitors who leave the site after viewing a single page has increased.

Additionally, it’s worth comparing eCPM, fill rate, and total revenue between the test and control groups.

Step 3. Don’t Jump to Conclusions Too Soon

The first few hours may yield unstable results due to a small data set, variations in traffic, and fluctuations in ad demand.

Therefore, define the test period and the criteria for scaling the format in advance.

A high eCPM alone does not necessarily mean the test was successful. If UX metrics deteriorate alongside revenue, the result requires further analysis.

Step 4. Record the Results

After the test is complete, compare the metrics of the two variants:

Revenue → Viewability → Time-on-Site → Bounce Rate → Overall Impact.

If the new format shows consistent improvement without a significant decline in user metrics, it can be gradually scaled up.

How to Run Tests in CTREX?

In the CTREX dashboard, publishers can set up split testing for ad placements without changing the website’s code.

This allows you to experiment with formats and monetization settings while maintaining your existing integration.

A proper A/B test isn’t about finding an ad that simply brings in more money. It’s about finding a solution that increases revenue without unduly compromising the user experience.

Start by testing on a portion of your audience. Measure the results. And only then should you scale up.