How to Split Test Ad Management Platforms

This is a short guide about split testing MonetizeMore’s award-winning platform with other adtech partners — what is it, how does it work, and what you can expect from it.

Split Testing Explained

Split testing, sometimes called an A/B test, 50/50 test, or multivariate test, is a transparent and agnostic way to prove, with real-life results, that MonetizeMore’s solutions are profitable even if you are paying us a commission over all your ad revenue, like BoredPanda (case study) and more than a thousand other websites worldwide.

Over the years, we’ve had publishers protest against significant revenue shares, only to find that we increased their revenues so significantly beyond our revenue share, creating a win-win scenario for everyone.  This is one of the ways we can prove it.

Our Split Testing works like this: 

We will give you a snippet of code that will split traffic between each side of the test.

• 50% of the traffic will run one ad stack, and the other half will run the other ad stack.  Visitors are kept on the same side of the test for the whole session so that the ad tech doesn’t mix user experiences or mess with buy-side treatment.
• Both sides of the test should run the same ad unit maps to have an apples-to-apples comparison.
• The snippet has been tested time and time again on many publishers, meaning it’s reliable and works well.
• We will work with your team to compile transparent reporting of each side.
• We’re an ad management platform, not an ad network.  We may need to acquire ad tech accounts or sub-accounts on your behalf.  For transparency in the test, we will also likely need additional tags or PIDs from any existing accounts that you already have.
• Test duration varies by publisher ad stack setup and usually lasts 2-6 weeks.
• Throughout the test period, we’ll compare the Session RPM of each side.
• At the end of the test, the version consistently achieving the highest Session RPM is by definition the most profitable, and therefore the winner.
• In some cases, publisher restrictions make obtaining the Session RPM impractical, so a different metric may be required.  We’ll go through that with you if this applies in your situation.

We consistently win tests across many verticals, even having beaten out competitors by over 200%.

The exact snippet you need depends on a few things and we want to work with you to keep it as clean and transparent as possible.  For these reasons, we need to know the answers to the following questions:

• What ad technology will be on the other side of the test?  (e.g. Adsense, AdX, Google Ad Manager, Header Bidding with Prebid, a custom wrapper)
  • If you can please provide us with a URL to that ad stack, even better.
• Are you managing the other ad tech in-house, or do you have another ad management platform provider?
• Are you loading the other ad tech directly on page, through a script manager like Google Tag Manager, through a WordPress plugin, or in some other manner?
• Do you have one or more full-time frontend/javascript developers on staff?
• Are you running any direct sold ad campaigns through GAM or PMPs, and if so, how much of your revenue is direct sold?  (if you don’t know what this is, then don’t worry)

What to expect?

• The snippet will load one side or the other, never both simultaneously.  Impact to latency and CoreWeb Vitals from the snippet is typically negligible.  Header bidding and traffic validation technology may add latency for some sites’ audiences, so please let us know whether you’re running either technology on the other side.
• At the end of the test period, it will be clear which version won, and you’ll be able to make your decision based on real-life performance in a fair way.
• To be 100% transparent and fair, we want to win the test, but in the rare case that we lose, it means we need to improve, which is something we are always eager to do.
• Always be weary of ad management platforms who are not transparent with data.  Every vendor should transparently provide you with the true impressions and revenue from each ad network, as well as stats like daily impressions, pageviews, ads per page, and breakdowns for mobile vs desktop.  For example, some vendors will not provide publishers with the raw GAM data and ad network data, effectively allowing the vendor to misrepresent what is actually happening in the test.  We’ve had publishers come back to us when they realized those vendors faked performance stats with “negative revenue shares” or phony direct campaigns during tests.  These vendors can’t keep this up forever, and revenue starts sliding afterwards.

When you’re ready, reach out to us and we’ll help you get started!