As always, an extraordinary analysis, thank you.

First of all, I wanted to point out that LeoThreads is the short-form content feature of InLeo (the app), which also publishes to the blog.

When I started reading and you mentioned that we would be surprised, I thought to myself, "No way, it's definitely going to be Liketu," and boy, was I wrong! 🤣

It catches my attention that rewards don't carry that much influence. It would be interesting to analyze that topic further, for example, the migration from one dapp to another due to the votes they distribute.

In that sense, the PeakD data surprised me, since they never vote for individual users, but rather for project curation reports. On the other hand, InLeo is the one that votes with higher amounts, but Ecency is the one that votes for the highest number of users. I know many users switch dapps looking for those votes, even publishing through them while continuing to interact on PeakD or Hive Blog.

Liketu used to vote quite a bit and had a good amount of HP, but they did a power down. It would be great to know if that power down led to a decrease in retention, marking the beginning of abandonment or migration toward others.

Sourse

There should definitely be a second part to this analysis. For instance, dapp churn, and how the migration between one and another looks. We have done onboarding across several dapps and have switched depending on the circumstances. Just as we change dapps during onboarding, the users themselves keep changing as well.

Regarding the InLeo case, it would also be interesting to check if the highest wave of churn happened right after their DHF-funded adoption campaign. Beyond the allegations made by some regarding the use of bots, it could be useful to know if massive onboarding inherently generates a high rate of churn. I believe we should aim for more massive onboarding but while maintaining a close, community-driven approach. It is a massive challenge.

I imagine the next analysis would shift toward community onboarding, rather than the dapps themselves. In the meantime, I am going to prepare our own data for Hive Run.