Sampling Strava

I’m an avid user of Strava, the social network for cyclists and runners. Strava has been very successful and lots of people are curious about how many members it has, how many active members, and how many are premium members, paying about $60 per year for advanced features.

I believe that Strava has about 8.2 million users, about 1.2 million have posted an activity in the past 24 days, and (perhaps the most closely guarded secret) about 192,000 are premium members.

How do I know? Statistical sampling.  Strava’s user numbers are sequential, and each user number can be used to see a member’s home page. For example, my home page is http://www.strava.com/athletes/5866.  See that 5866 at the end?  That’s the key.  I was the 5,866th person to sign up for Strava.  If you sign up today, you’ll get a number around 8,200,000.  Over 8 million people have signed up for Strava.

Because virtually every user number below 8,200,000 links to a real person (try it – it’s fun!), I was able to set up a statistical sample of Strava’s users.  I chose a sample size of 384 users to get a 5 percentage point confidence interval on a 50% occurence at 95% confidence level, then I generated 384 random numbers between 1 and 8,200,000 and started entering them into Strava.

This set me on a global adventure, visiting (briefly) with wonderful, adventurous people around the world. I’m not a statistician but I tried to keep this as scientific as possible.  I recorded my results on a Google sheet that can be viewed here.

Active Users and Premium Users
The biggest question I was trying to answer was how many people were paying for Strava.  My sample revealed 9 premium members out of 384 sampled members, or 2.3%.  Extrapolating that over Strava’s 8.2 million users yields an estimate of 192,188 premium members.  My 95% confidence interval is +/-1.5%.  That is, I’m 95% confident that the actual number is between 65,000 premium members and 311,600 members.

I was also interested to see what percentage of the members were active. I defined active as having posted a ride or run in the last 24 days because that’s what showed up graphically in the home page – Monday, March 9 through Wednesday, March 11 plus the previous three weeks.

I found 58 active members out of 384 sampled, or 15.1%. Considering that it’s still winter in the Northern Hemisphere, that’s pretty impressive.  Extrapolating to the larger sample, Strava has 1,238,542 active members by my definition, with an 95% confidence interval of 3.58% – I’m 95% confident that the actual number is between 944,640 and 1,531,760 active users. Incidentally, every premium member was active.

Why does this matter? Many of us are passionate about Strava and want it to be sustainable.  The company raised $18.5 million in October 2014 and is not yet profitable.  This analysis suggests that Strava’s revenue from premium memberships is around $11.5 million per year. Besides, Strava’s free service is pretty great – the company hasn’t started pushing hard on existing users to switch to premium. If it can eventually convert many of its growing number of active users to paying customers, it stands a good chance of keeping its product free from intrusive advertising and clutter.

Next Up: More from the Strava Sample
I learned many more interesting things from sampling Strava’s users, and I’ll follow up in a subsequent post highlighting Strava’s astonishingly international scope and the U.K.’s obsession with Strava.

Note: I have many friends who work at Strava but I consulted with none of them regarding this analysis. I hope that they find it interesting and are inspired by the breadth and passion of their remarkable users.