Hi Creators,
Today, we’re excited to share an update to thumbnail personalization that lets you test new thumbnails while still sending most traffic on your existing winning thumbnail.
Our experiments showed a +0.19% increase in Qualified Play Through Rate and 21% fewer impressions going to a non-winning thumbnail.
How we’re improving thumbnail personalization
Previously, starting a new thumbnail test would reset the system and split all traffic across thumbnails evenly. This could potentially impact your QPTR in the short term while the system finds a new winner.
Now, when you start a new test that includes your existing winning thumbnail:
- Your existing winning thumbnail will still receive the most traffic.
- New thumbnails get enough impressions to measure QPTR and then traffic automatically adjusts within hours based on performance.
This approach ensures new thumbnails get enough exposure for effective testing without sacrificing your current winners’ performance. You’ll also know sooner how new thumbnails are performing.
In the example below, suppose you started testing your existing winning thumbnail and 4 new thumbnails at 7 PM. Personalization will now still send most traffic to your existing winner while testing the 4 new thumbnails (right) instead of splitting traffic evenly (left).
Before today’s update, new thumbnail tests split traffic evenly (left). Now, your winning thumbnail keeps most traffic while new ones get tested (right).
What you should do
Whenever possible, we recommend that you always include your highest QPTR existing thumbnail when starting a new thumbnail test.
This way, we’ll continue showing most users your existing winner while sending a small percentage of traffic to test your new thumbnails. The total number of uploaded thumbnails remains at 5.
For more best practices, check out our documentation.
Let us know if you have any questions or feedback in the comments below.
Thank you!