It's been quite a while since I looked into user retention on Hive. This is a topic I've written about many times over the years, you can find examples of my past posts here. Please note that my statistics focus on authors, for the simple reason that it is easier to verify that most are 'real individual users' compared to alternative measurements.
Brand New User Retention Collapsed In Recent Years
Actually, it seems fairly unfortunate that my last look at overall new user retention was all the way back in 2023, because it appears in the meantime I missed this:
In mid 2023 monthly brand new user retention dropped sharply from a rate that had been consistently above 35%, to an all time low of 16.4% in December 2024. Since then, it appears to have been recovering, but is still well below our healthiest days.
Cohorts
Another way to look at user retention is through the lens of cohorts. I first wrote about Cohorts on Hive in 2022 and revisited the topic in 2023.
The following chart shows active authors on Hive split into cohorts based on when they created their account. Pre-2020 accounts are in yearly cohorts, all accounts created from 2020 on are in quarterly cohorts.
Next is posts per month, split by the same cohorts.
Post rewards split by cohorts:
The proportion of author rewards going to each cohort:
You can see above that the pre-2020 cohorts are collectively receiving over 40% of author rewards despite being a smaller group now. The 2016 cohort gets an especially disproportionate portion of rewards, they are now a group of only about 160 active authors.
Circling back to the original topic, another way of measuring new user retention, by comparing the monthly change in number of users in the newest cohort, shows the same collapse in new user retention in 2023/2024.
Conclusion
Brand new user retention took a severe dip starting from mid 2023, reaching its lowest point in December 2024. It has partially recovered in 2025.
I plan to follow up and investigate what might have caused the drop in user retention, the recovery, and what the impact was on user retention overall, but I have not determined that yet.