What is a Contextual Bandit?

With the increased development of Algorithms these days, testing has become somewhat of a minefield with multiple approaches to tackle different problems.

Contextual bandits are algorithms that rely on incoming context data to make decisions in real-time. Bandit algorithms allow websites to constantly load different variations of a webpage and adjust based on visitor behaviours.

How is Contextual Bandit used?

Different types of people react differently to different materials, which is why it is impossible for there to be a single losing variation.

Let’s say you own a fashion retail site and want to engage with your female users. A/B testing may show that promotions are best for sales, but if the majority of your visitors are women, it could lead to false conclusions about how well current marketing strategies will work.