A/B testing analytics for React Components


Hi There,

I have an alpha library/SDK, and barebones analytics service that allows developers to quickly, easily, and robustly deploy A/B tests into their React Apps. It's been running successfully on some applications run by friends, and some of my other projects, so now I'm starting to test the water with some marketing.

This is my first stab at communicating what it does and why it's useful.

  1. If you're a React developer do you get it?
  2. If you're not but you know React developers that wan't do to A/B testing do they get it?
  3. Is it at all compelling to non-developers?

Any feedback welcome.

Thank you!

J x


  1. 2

    This is pretty cool. Not sure how it is different from existing solutions. Maybe you can provide a library of UI widgets to facilitate easy integration for A/B tests?

    1. 1

      Thanks for checking it out! I think the differences I've received from testers and surveyed teams are:

      1. Generalist analytics platforms are difficult to navigate on the analysis side - not tailored specifically to A/B testing
      2. They leave a lot of the multivariate testing math up to you - and people don't have the resources
      3. I haven't encountered any SDKs that are tailored to React Component concepts (they very well might exist though! Which existing solutions do you think you'd recommend to friends at the moment out of interest?)
      4. People spend time and money on their own solutions

      It's an interesting point about UI widgets, but I think we'd want to leave the components up to user-land. We're just an analytics tool!

      Thanks again for checking it out!

      J x

  2. 1

    At the company I work, we did something like this but under-the-hood we are using optimizely.

    The message is super clear.

    What I don't know (I don't have lots of experience with A/B test) is:
    1- How are you identiying the test. I ask this because I don't see any experimentId or things like this.
    2- Does it support or will support... % user distribution between test variations?

    I like the idea so much! Congratulations 🎉

    1. 1

      Thank you very much, for the compliment and taking the time to comment!

      1. At the moment I let the user determine the ID, I've thought about putting some programatic checks in the SDK to defend against multiple tests having the same ID. It's a bit fragile, but one of the pieces of feedback I got about setting up tests was the friction of going into a dashboard to set up a test. I'd like devs to be able to configure a test entirely through code. Potentially I could try a CLI at some point...

      2. User distribution will come if I can afford it - it requires more calculation and database activity on my servers - if customers really push for it - it is something I'd like to add!

      1. 1

        yep! I think that would add a lot of value! congratulations :D

Trending on Indie Hackers
Indie Hackers is now an invite-only community 74 comments The Challenge: $10,000 MRR in 30 Months 41 comments Do you *like* Twitter? 33 comments 26 B2B Cold Outreach Templates - all for free... 🤝🏾 16 comments I founded CircleCI (valued at $1.7B) and Darklang. AMA! 11 comments 29 SEO lessons after growing my startups from 0 to $2 million in revenue & 400K/month organic traffic. 6 comments