How Symmetry grows

A/B tested one extra screen at the end of onboarding where the user commits/signs that they will work out, against control. Tracked in PostHog.

Built an internal dashboard tracking every UGC video, segmented into formats and sub-formats. ~80K videos in a year from 500+ accounts. Example winning format: iPhone lock screen showing a personal notification (e.g. 'your girlfriend leaving you') with a Symmetry notification below saying 'go to the gym and work out'; one such video got 17M views.

A/B tested two completely different designs of the first screen after onboarding (where users see their workouts) against control.

Start from a problem, not a solution: What problem am I solving? How do I know it's real? Then list solutions/hypotheses, estimate how much each moves the target metric, define guardrail metrics, and A/B test control vs new version for at least two weeks before analyzing. Maintains 'Evelyn' (Experiment Velocity Engine List) database of experiments sourced from user interviews; product experiments in PostHog, paywall tests in Superwall. Evolved from building what founders wanted (terrible) to building user requests (mediocre) to this.

Founders were fitness YouTubers and launched to their YouTube community.

Gym tracker built for the Spanish-speaking market (mainly Spain and Mexico) rather than competing in the US.

Tried influencer marketing after the YouTube launch; no details given.

Prioritize screens that affect most users radically (paywall first) over low-reach features like dark/light theme. At Symmetry the exercise search screen strongly affects activation and retention, so it gets experiments; the post-login workouts screen does not.

Every 90 days ('warmup') set marketing and product objectives with metrics reviewed weekly or daily. Early products should set very ambitious OKRs (e.g. +10% conversion in 90 days is easy for a new app). For a gym tracker the key metric is activation (tracking a first workout), since retention follows activation.

Rank for every muscle group and every exercise, levels per exercise, Strava-like social feed where users share workouts, and streaks visible to friends.

Nearly half of company revenue goes to creators, so they are incentivized to produce more content and stay with the app long-term. Uses ARPU per download to decide what can be paid for downloads. Exact comp structure (per video, per view, rev share) not stated.