Have you noticed that Rihisi changes over time? Maybe you’ve even seen a change that quickly disappeared. Here you can read why and how it is happening!

At Rihisi, our teams run numerous tests and experiments to understand our users better and improve the experience for tutors and students. In this article, we want to give you an overview of our experiment culture and provide some real examples.

Before we make changes on Rihisi, we want to ensure that these changes benefit tutors and students. That’s why we run experiments, in which we make changes for a select group of users so that we can compare their experience with users who haven’t experienced the change. After assessing the results over a period of time, we analyze whether the experiment had a positive effect. If it did, we’ll make the change for the rest of our users. This is how we continue to grow, and improve your experience on the platform.

Most of our experiments involve adding or changing features on the platform, such as the Rihisi classroom, the Chat function or the Calendar. We normally decide to run these experiments based on data we've collected or feedback from tutors and students. Most of the features you use and love today started as experiments. Of course, not every experiment is successful. In these cases, we "kill" the experiment and things go back to how they were for all users.

This normally depends on the type of experiment we’re running. Some experiments happen on the tutor side, meaning a select group of tutors will experience the change. Other experiments happen on the student side, meaning you might only notice the change when teaching specific students. Our big experiments normally affect 50% of our users, while smaller experiments might only be conducted within a specific group. Whether the experiment affects all users or just a specific group, those who experience the change are chosen at random to avoid bias.

In most cases, experiments last between 3-5 weeks. This allows us to gather enough data to be certain about whether the experiment was successful. We might run some tests for a longer period of time. We do this when there’s reason to believe the short-term effects might not accurately represent the long-term effects.