Yup. Hard truth.
Discussion
brutal looking chart
Well not entirely if a large portion of those are automated profiles.
I'd love to see retention numbers for 'real' users (whatever that is)
What’s your cohort for W0 including? This makes no sense at all.
It includes any new pubkey we saw publish their first event in that week. I'm happy to hear ideas on how to better tune that.
That’s a tough one. There’s likely a lot of testing going on. Like I know we make keypairs all the time to test. Probably many other cases as well.
First thing that comes to mind is to widen the band of w0 to m0. Like consider anyone that doesn’t stay for at least a month a non-user and see what that drop off rate is. It’s a blunt idea. Hoping your convo with nostr:nprofile1qqs0dqlgwq6l0t20gnstnr8mm9fhu9j9t2fv6wxwl3xtx8dh24l4auspzamhxue69uhhyetvv9ujuurjd9kkzmpwdejhgtcpz4mhxue69uhhyetvv9ujuerpd46hxtnfduhsf92fn2 is more fruitful.
Agree. Assuming active is pub key creation + 1 event is probably too slim.
Event threshold + 1 month stay to be considered a real user.