Hi Samuel.
Nice job! Twitter follow and upvote from me! I wanted to do a project like this some day 😉
Main point for me was to turn unstructured, yet-unrecorded data into valuable insights, and do it at global scale.
Some thoughts drew me away from making this app:
1) Something I would call "entry points" for this app. It should be released as iPhone and Android app and preferably some public JSON API to record events.
I assumed I won't be able to validate the idea if I won't provide enough "entry points" (apps, web, API) for my end users. They cannot be blocked with single entry point (web UI). Of course you can start small and grow organically (design iPhone app when users really need it).
2) Events (check-ins) are rarely "single points". They can have extra meta data (duration, location, whatever). If I track events with sentry or with GA, it collects many many extra data for me. If I would measure something with my app, I want to have extra dimensions and analyze how it affects results. It makes both UI and backend significantly more complicated.
I noticed that I could do some low-level open source (like charts library) or SaaS (like scalable BaaS for collecting events) instead to enable community to create useful products.
Again, you can start with basic events (check-ins) and expand later.
3) User experience issues, especially human computation and other trends. I think that only some small fraction of geeks want to explicitly track some events. I think that conducting experiments at scale should happen like in Duolingo or Recaptcha. I know that this was not the main point of the app. I just thought that I can do it better with gamification or advanced UX (like Duolingo guys).
I don't mean that idea is wrong. Duolingo-as-an-Experiment simply questioned my first, explicit and straightforward approach (on many levels) and drained my initial mental power to write the app.
@szymon_korytnicki thank you very much for this comment. Very interesting, indeed. It turns out that there is an API available (for some "connector"s I've built manually) but that is (on purpose) not documented. The main point being validation of the idea. I don't see this tool as a replacement to a GA, Sentry & cie: it should be simple on purpose, to help focus on the more essential, I believe.
The user-experience issues is a very interesting point. You are right, we are only a small fraction of people (I'd say it's "a few geeks" but also a lot of product people).
> I think that conducting experiments at scale should happen like in Duolingo or Recaptcha.
What do you mean by that?
Pushletter
LetsExperiment.today