Integrating with Firebase A/B experiments

  • 30 August 2021
  • 3 replies
  • 312 views

Badge +4

Hi everyone,

 

I recently started a premium subscription for my app on Google Play Store.

I started using RevenueCat to handle purchases, but I struggle to integrate it into my A/B tests that I run via Firebase Remote Config.

 

Here is the summary of the setup. The  users are presented with two choices of a subscription screen (which is recorded as a Firebase event). When a user completes a purchase, the app gets notified correctly and fires another Firebase event (along with enabling premium content for the user).

So this part works fine and I can analyze the relative performance. The problem is with the cancellations. Since a user can cancel the subscription and never open the app again, I have no way of processing them reliably in my app code. So I need a different way of receiving these events from RevenueCat and into Firebase.

 

I did not find any documentation on that. So far, I manually copy cancellations from RC console into the spreadsheet, but this is time consuming and unreliable. I would greatly appreciate an advice on how to properly handle this!


3 replies

Userlevel 5
Badge +9

Firebase doesn’t have a REST API for their analytics, which is used for their A/B testing, so I don’t think there’s a way to integrate any outside services. (you can only run Firebase A/B tests based off in-app events recorded through their SDK).

 

What I do is use Firebase Remote Config to control the RevenueCat Offering that each customer sees. I can then do some analysis inside RevenueCat charts and Customer Lists on which Offering is performing better.

 

I had this setup in place before the RevenueCat Experiments feature was in place. The benefits of Experiments is that the analysis is done for more me using more complex models than I care to build myself.

 

Curious to hear if anyone else has done any creative setups with Firebase A/B testing and RevenueCat Offerings?

Badge +7

I used Firebase A/B for pricing experiments but the maths was too hard/complex for me to confidently analyse and so switched to RC experiments instead. Much nicer to see the needles moving about for someone simple like me. 

 

I’m also unconvinced that Firebase tracks all our in-app purchases correctly (wild differences between it and our RC dashboard). 

Userlevel 5
Badge +9

I’m also unconvinced that Firebase tracks all our in-app purchases correctly (wild differences between it and our RC dashboard).

It seems that Firebase Analytics only tracks in-app purchases that are detected client-side, which can mean missing trial-conversions and renewals. It also seems like they sometimes mix up a free trial for a paid subscription 🤷‍♂️ Regardless, I wouldn’t recommend Firebase Analytics for any subscription revenue tracking. Mixpanel or Amplitude would be better since you can forward renewal events server-side (or through RevenueCat integration) as they happen. 

Reply