Seeing Farcaster DAU chatter on my feed today and it's reminding me of the time I was put in charge of calculating user counts for a public b2b saas company.  Active user counts were a big deal at the company.  Founder and CEO talked about user counts a lot. Showed up in every slide deck, on our website home page.

Cohost /gmfarcaster livestream Farcaster news show 🎙️💬 Founder Tenger Ways, bringing joy into work with DevOps practices 🤗 mom x 3 💗

The problem: it took 5 days to calculate.

I was put in charge of business intelligence and my first task was to "automate user counts".

CEO wanted a dashboard.  But every earlier attempt led to "our data sucks" "we can't trust the dashboard"

Because we were reporting to Wall Street in our quarterly earnings report, we had to make sure we didn't include users from demo portals, test/QA clients, or clients that hadn't yet gone live.  But we didn't track client status anywhere in a database; info was in various spreadsheets and in the COO's head.

What about usage though? If a user never logged in, should they be counted as active? Well, since we charged by the seat, not by usage, it didn't matter whether the user logged in or not.  Well, at least not to Wall Street.  But what about to the UX and Product people?

I came to realize something as simple as user counts means something very different to different audiences.  

Wall Street = user counts as proxy to paying customers
UX/Product = actual users logging in and using the software
DBA/CTO = all users to model future infrastructure needs and cost

Depending on who you were, active user meant something different.  And you only cared about your definition.  You were too busy and myopic to even realize there were other definitions.  Everyone called their definition "user counts", which is why no one trusted the data or a dashboard

Ultimately we did solve this problem and created a dashboard for Active User Counts that people could trust, but the hard work was less engineering and data science, and more social and organizational - aligning people around a common language before looking at data

Easy peasy, I thought.  My engineering mind thinking it would be as simple as "select * from users where status = active"

Only after weeks of research did I realize why it was taking so long to calculate.  No one had ever defined exactly what it meant to be an active user...