A/B Testing is Here to Stay

According to MediaPost.com, having been around for a long time, A/B testing couldn’t be fading away, thanks to the advances made in the artificial intelligence, it could be going to back of the line. Recently, at the eTail East conference that occurs annually and exhibition in Boston, the hall for exhibition was packed with companies that were promoting numerous versions of customer targeting, data analytics, plus most prominently, different AI-driven promises results.

MediaPost.com interviewed Sentient AI’s marketing director, Jeremy Miller to kick around the potential future and current status of the A/B testing old world concerning the AI new world. Miller said that in traditional formats of A/B testing, you have your control Versus an experiment. You run the experiment on your traffic; whichever design that’s going to perform better is the one you’ll use, which is the true and tried practice.

People have discovered six out of seven tests do not end in a positive result, so one has to put a lot of resources and energy trying to determine how you could increase conversions by the use of A/B testing. The conversion-focused product of Sentient known as Ascend allows marketers to test all their ideas at the same time instead of in a sequential, linear way. It’s having evolutionary algorithms and mimics the evolution process, taking all ideas plus does continuous optimization.

Two to three years, people started taking advantage of AI to help with testing methods and Miller says A/B testing would not go away anytime soon. He’s confident A/B testing is affirmatory. If you are looking for confirmation of a hypothesis, Miller said an A/B test could still provide certain values. You could be having a strong hypothesis that you need to check it it’s accurate, then A/B is a process of doing that. It is already decided what you think is right and you are just invalidating or validating.

When you are exploring, you are navigating toward the greatest possible conversion. You are exploring without limit or reservation to the mountain top to reach the highest peak versus only this local peak. An A/B test will get you that. You could be lucky with an A/B testing, but you did not make room for the probability that something else might have allowed it become better; the Sentient’s AI will eliminate the issue of wishing to get lucky.