Lessons From Building a 70 Person Growth Team — Jason van der Merwe, Strava

The product analyst found that the first upload is the most important new-user action, yet a funnel analysis showed a big drop-off. The small growth team ran onboarding tests on usertesting.com and watched them together; users reached the record screen and asked 'What is this? What am I supposed to tap?' because the buttons mimicked a TV/DVD remote (round red record, square stop). They added text to every button, using short words like 'GO' that localize well into small buttons.

The growth team's first-ever experiment added the ability to set a profile photo in the onboarding flow.

Strava's activation metric: upload one activity (above a minimum distance, so a test tap doesn't count) and follow one person within the first 7 days. It's hard because it requires going outside, unlike metrics you can hit from bed. Most early growth work focused on reducing friction to that first upload.

Activity sharing already generated several hundred thousand link shares a month. The early acquisition team worked to raise the number of shares and improve conversion of visitors on the receiving side, as a referral loop built on user-generated content.

An engineer (Annie) ran it on Experiments Day: if someone followed you and you saw their profile anywhere in the app, the button said 'Follow back'. It had been deprioritized because data showed most Strava relationships were already reciprocal.

Once a quarter (mostly when in person), a day where anyone on the growth team could run whatever experiment they wanted without buy-in from an engineering or product manager. Rules: don't remove a subscription upsell, don't crash the app, expose at most 10% of users. Teams still collaborated across design, analytics and engineering, and often ran ideas that had been deprioritized. Done less recently.

A weekly 'experiments weekly' meeting presented every experiment the growth org ran. It worked at five people (debate, follow-up test running within 2 days), but at ~70 invitees only leads spoke. After one team started its own small version with great brainstorming, the big meeting was cancelled in favor of per-team meetings. To keep cross-team learning, the director reads all recent experiments on flights and sends pre-written Slack questions.

While working on the feed, the growth team built a service for feature education: coach marks, tooltips, popups and in-app units, plus push and email, aimed at new users and win-backs/resurrections rather than daily power users. For about two years only growth used it; later every team adopted it. The activation team is now running a negative test that removes all feature education to measure its overall impact.

On the activation team, tested the photos in the logged-out welcome experience ('front doors') before signup, cycling vs running imagery. Findings were applied to the brand and carried into App Store assets.

Growth partners with the copy team to run copy tests, e.g. 'Continue with' vs 'Sign up with' on login buttons. Brand photography placed in product settings tends to lower conversion (theory: it feels like an ad or distracts), so copy is prioritized over imagery in product flows.

Aim for four nines (99.99%) crash-free sessions and service availability, and use Apple's and Google's native review prompts to ask engaged users to rate the app, since store ranking weighs five-star ratings.

Store screenshots were rarely changed because they couldn't be tested or attributed; even with native testing, novelty effects fade after a couple of months. Focus instead on making it very clear what Strava does, with app preview videos.

Used a service that builds a web page mimicking the App Store listing, optimized it for click-throughs, then shipped the winning creative to the real store.

Apple regularly offers Strava features in specific countries (e.g. Germany) and asks for special assets, which Strava provides.

The first growth team was effectively an activation team (new-user activation is where most early wins are). It expanded to acquisition (supporting paid, referral loops via activity sharing), then absorbed the subscription/monetization team (upsell conversion, friction), then a communications & personalization team for notifications, email and in-app messaging, inspired by Pinterest's growth comms team. New areas start with one person running experiments; if it works, invest more. Teams are cross-functional with growth and product marketers embedded, and new teams are told to just run A/B tests for the first months to learn the idea-to-evaluation loop.

The product analyst turned a SQL query he re-ran for every test into an internal experiment reporting framework ('One Ring') with a Tableau dashboard; nearly every experiment reports its metrics there, run overnight. Teams keep a small set of measured (not modeled) outcomes. More recently they model how moving an output metric (often week-2 retention, e.g. for win-backs) changes WAU over 12 months with decay built in, and use long-term holdout groups to guard against models trained on their own outputs.

Because employees are daily power users, Strava built a switch that makes an employee's profile free; most subscription team members stay in that mode to keep empathy for free and new users.

Before the mobile app, Strava's founders drove a van to local cycling races, got cyclists to come over and plug in their Garmin devices, and uploaded the rides to the Strava web app for them.

To expand from cyclists to runners (competing with Nike Run, Runtastic, Runkeeper), Strava launched a separate Strava Run app alongside the ride app, then combined them into one app around 2015.

Strava's marketing team grew local communities, went to events and built relationships with professionals (e.g. Chris Froome, Molly Seidel), whose activities fans can view and compare against.