A/B testing, also known as split testing, is comparing any two or more combinations of offers and/or landing pages to see which one performs better. To put it simply, the one that gives a higher conversion rate (CR) and earnings per click (EPC) wins.  A/B testing allows you to maximize your existing traffic. While the cost of acquiring paid traffic can be huge, the cost of increasing your conversions is minimal.

Our ML system analyzes traffic to make a recommendation.  This is how it works for Track Revenue.

Say a direct-to-to offer campaign gets 1000 clicks a minute. For each click a request is sent to the ML API that asks, “which offer should I use?” for the campaign’s active offers. At first the ML API doesn’t know anything because it hasn’t been trained, so it returns a random recommendation. Eventually a conversion happens. That gets sent to the ML API where it is joined with the original click data. That tells the ML algorithm that the click, and the data associated with the click (region, country, platform, device, etc.) resulted in a reward/conversion.

The more conversions happen, the more the algorithm is “trained.” When another click comes in the algorithm looks at all the click’s data and the reward/conversion information that has been collected, then makes a recommendation about the best offer for the click.

The final piece of the process is the exploration pool. This basically means that 20% of the time the ML API will return a random recommendation. This is to ensure that all options have a chance to become the best offer and any options that increase in performance over time are not discounted.

The entire process occurs instantaneously within milliseconds, which means that Track Revenue is optimizing conversions and revenues in real time. Results speak for themselves and our ML auto-optimization has proven to boost customer’s EPC by an average of 38%.

Give Machine Learning a Free Trial today and don’t miss out!

Track Revenue Team