We used Genetic A/B Testing to increase Website Conversion by 300%

At FeedHive, we love to do A/B testing! Especially on our marketing site.

But we got really tired of planning, setting up, and executing experiments, not to mention collecting the results and deploying them - often in combination with the results of other A/B tests.
Tedious and time-consuming.

So - we started using Genetic A/B Testing instead!

What the h... is "Genetic A/B Testing"? 🤨
Well, if you don't know - I don't blame you.

It's a term we came of with to describe a system we created, which builds on top of regular A/B testing, but takes an "evolutionary" approach instead.

Eventually, we ended up increasing our website conversion by almost 300%!
Waaay faster than we could have done under normal circumstances.

Let me break down how it works:

We line up a queue of variants (or mutations, if you will).
60-80% of incoming traffic gets directed to the "main" variant, and the remaining 20-40% gets directed to the "experiment".

Here, we roll a dice and send the visitor to one of 4 active variants.
These 4 variants now "compete" in order to take over the "main" variants position.

They have 10 days to do so.

If they manage to beat the "main" variants conversion rate, they'll automatically get promoted and become the new "main".
Otherwise, they go "extinct" - which means they get removed and replaced by the next variant in the queue.

We implemented this using AWS, CloudFront, and LambdaEdge.
The whole thing actually started out as one big experiment, but we were so happy with this system that we still use it today 💪

If you're interested, I'm breaking the whole concept down into details in this YouTube video:
https://www.youtube.com/watch?v=eJDFyIKIBKM

Also - I'm interested in hearing, what do you think about this idea?
Have you tried creating something similar for your products?