In my last post it was demonstrated how between 30 and 40% of Hive authors consistently go inactive every month. Among those, about 6-10% return to make another post, but most are gone for longer or for good. Although this is disheartening, it was also shown that this is a marked improvement over user retention in the days of Steem, when it was normal for more than 40% of authors to leave each month, at times it reached 70%.
The most worthwhile insight from that data was the following: lower Hive price and lower author rewards have little effect on user retention. Authors leave Hive for some other reason(s) than rewards or Hive price - they leave in the same proportion each month regardless of price.
Yet, we all know that Hive user activity is higher when price (and thus rewards) are higher. This can be seen by simply overlaying a price chart with monthly active authors. I did this with @coingecko charts crudely below.
Hive post fork:
Steem pre fork:
Without going into any math, it is clear that there is a relationship between active authors and price.
Since we know that the number of authors leaving is unrelated to the price - it happens at a fairly consistent rate especially in post-fork Hive, new authors joining must be related to the price. That's not to say that a high price necessarily causes an increase in active authors or the reverse - but the two are linked in some way.
What is More Important, Bringing in New Users or Increasing Retention?
I want to go beyond just looking at our current user retention rate and look at what the possibilities would be if we improved user on boarding and user retention.
Imagine a scenario where we had 20%, 40% or 60% more new users join the network every month since the fork, but with the same rate of attrition. This may be the result of improved marketing, improved user account on-boarding or perhaps just increased attention on Hive during crypto price boom. How many active authors would we have today?
Clearly, we'd be doing much better with more users joining. However, notice the diminishing returns. The first +20% gets us much more than the second or third. Ultimately, if we can't sustain the increase in new users, our high attrition rate brings us more or less back to where we are today.
How about if the attrition rate is reduced instead? In this case, we will assume that new users per month is unchanged.
A 20% reduction in attrition gets us basically nowhere, yet a 50% reduction mades an enormous difference, pretty much as good as +60% more new users consistently every month. I suspect that this is a critical mass effect - like a snowball reaching a point where it is only gaining more and more mass.
While we're dreaming, how about if we achieved both an increase in new users AND a reduction in attrition?
The combined result of both increased new users and increased retention would be substantial, sustainable growth.
Going Exponential
In creating the above charts, I implicitly made the assumption that new user rates are not proportionate to existing user activity. This assumption simplifies the math, but is likely false and makes the charts above conservative. If new users join at rates that are proportionate to existing activity, then that could be an exponential factor which has been ignored. @elmerlin has a brain-value model where price is linked to user activity, and it is not a controversial idea in the Hive space that there is some price impact and growth factor contributed by the existing network of users.
In reality, there is a positive correlation between new/returning authors and active authors of 0.542. This is considered a moderate correlation, and if the relationship is causal then the more active authors the more new users we can expect. This can be accounted for in the model by adjusting the number of new users who join based on active users and it alters the results... to the point of insanity.
This fails the smell test. It does not seem likely that reducing the attrition rate by 20% would bring us to 1.5 million active authors in just over 2 years. Not completely implausible, but more likely taking the correlation at face value overestimates the causal relationship between active authors and new authors joining/returning, assuming it exists at all.
Here are some results when we reduce the relationship by a factor of ten, which may represent a more realistic result. Again with a -50% reduced attrition rate Hive would have an easy time achieving critical mass.
Let's go back to user onboarding again, including the same factor of a relationship between active users and new/returning users.
With higher new user rates, we still grow much faster than we do now. With at least a 40% increase in the rate of new users joining, it is equivalent to reducing our attrition rate by 50%, however it may fall back to similar levels as today with time.
Finally we will look again at what a combined approach would do, assuming a weak causal relationship exists between active authors and new joiners.
Conclusion
Increasing user activity on Hive will require improving our user retention AND increasing the rate at which we acquire new users. An increasing rate of new users will grow our network faster initially, but is unlikely to be sustainable without also improving retention.
If there is a causal relationship between current user activity and the rate at which we attract new users, as many believe there is, then the more improving user retention will help to grow the network, and the easier it will be for Hive to attain critical mass.
It should be a priority for those of us with a stake in the project to identify the reason why 30% of authors become inactive each month, and to solve the problem that causes it.
I got some good comments on my last post relating to RC's and also how catfish accounts influence the statistics, I plan to address the issues therein in another post soon.