HOW PREMIUM PUBLISHER 9GAG USED A/B TESTING TO SELECT A HEADER BIDDING WRAPPER SOLUTION AND IMPROVE MONETISATION

As an organisation used to testing and learning, when it came to selecting a new wrapper solution to improve programmatic monetisation, 9GAG opted to run A/B tests to find the best partner.

9GAG constantly re-evaluates their stack and was keen to understand whether other wrapper solutions might drive better results for them compared to their incumbent wrapper.

9GAG selected PubMatic’s Prebid-based OpenWrap to test against their incumbent and one other industry leading wrapper.

Despite the obvious benefits of testing wrapper solutions before implementing them, it is not standard practice amongst digital publishers.

9GAG deployed a robust and randomized A/B testing methodology to compare header bidding wrapper performance. To ensure the tests were unbiased, objective and did not compromise ongoing monetisation, 9GAG utilised a previously developed randomized dice roll script to create test buckets and control the size of each bucket.

This dice-roll logic mechanism relies on two IDs—a Range ID and a Bucket ID—implemented by 9GAG, to ensure that all wrappers were given access to the same volume of traffic. A randomly generated Range ID is assigned to each user that visits a page, with cookies being used to determine repeat users.

A bucket ID is assigned to each wrapper with bucket range set as identical across all buckets. This setup gives each bucket the same probability of being assigned a user. Each bucket gets a very similar number of users assigned.

9GAG ensured the testing environments were fair and transparent for all test participants. This meant enforcing the same timeouts, bidders, and percentage of randomized global traffic. They also migrated from their wrapper provided consent manager platform to a neutral third-party provider.The tests were conducted over several months.