I didn’t see any existing topics related to this, so I wanted to start a convo around how others are integrating Optimizely with Amplitude.
Our current integration is as follows: When Optimizely detects a new user qualifies for an experiment, determines the variation to bucket them into, and then effectively enters said user into the experiment under the chosen variation, an event is pushed to the datalayer with the experiment name and variation. We then send this information to Amplitude as a user property. This is pretty similar to the Amplitude support documentation on integrating the two platforms. We built our custom integration before this documentation was released.
One interesting thing I’ve been seeing is that accordingly to Amplitude, some users are being grouped into different variations within the same experiment. I’ve ruled out merged profiles as a reason for this finding, so I’m thinking this has to be an issue on Optimizely’s end and I’m working with their support team on this now.
Another thing I’m considering is sending a custom event to Amplitude when a user enters an experiment, in addition to the user property. I think this will allow me to build more accurate behavioral funnels, with experiment entry being the beginning of the funnel.
Some questions I have for other folks who have integrated the two platforms:
- What does your integration look like?
- Have you validated Amplitude data as it pertains to Optimizely experimentation performance with the data seen in Optimizely? Did you see parity between the two or do the two platforms show different findings?
- How do you use Amplitude to enrich Optimizely findings?
- Any other tips/tricks that has made the Amplitude/Optimizely integration more valuable?
Looking forward to hearing from the community on this one!