How BeHard grows

UA managers ran TikTok influencer campaigns. While daily downloads are small, attribute each video by the traffic peak: note the post day (e.g. Wednesday), then that day's downloads and sales. At high install volume this becomes nearly impossible. Once a creator performs, sign a long-term deal. Discount codes capture only ~10-20% of influenced users; measure the code rate on early campaigns and model future ones from it.

Challenges were already popular: many people posting on TikTok and searching Google and the App Store. First campaigns aimed to capture a slice of that traffic via organic (ASO, TikTok) plus Apple Ads. Phase two was influencer marketing; phase three paid ads to reach $1M ARR. Guest warns Apple Ads is hard to scale profitably in competitive categories, and ASO alone is rarely viable unless validated before launch.

Founder made the first screenshots himself in a couple of days. Designers then spent 2-3 weeks on 'perfect' screenshots, and a hired agency produced five variations; both performed worse than the founder's set in store tests.

Dump every idea into a Notion table. Score interest, expertise and opportunity; a formula averages them and you pick the highest. Opportunity gets a point each for: demand, revenue proof, a balanced competition level (too little = no market; only one successful product is a red flag), retention (solves a real problem or truly entertaining) and timing (new platform features, new models, social trends). Expertise scores 1-4 on: can I build it, can I market it, do I understand the product, can I monetize it. Research with Foreplay, AdSpy (competitor Meta ads), Sensor Tower, Appfigures, Google Trends and TikTok view counts for keywords; validation usually takes no more than about a week.

Pick the top-scored idea and build it in a week, removing anything that would take too long. BeHard's first version was a bare challenge list (first workout, second workout, 10 pages). Use AI coding agents (carefully with autonomous ones); an 'agent factory' could test many ideas before investing human time.

Treat launch as the first experiment. If it isn't profitable, decide if it's a distribution or product problem (get traffic to the current version before improving the product). Find the bottleneck (CAC too high, low conversion, low LTV, low ad CTR), design an experiment for it, run it, analyze. If profitable, scale until it stops being profitable, then repeat. Start pricing by copying competitors. To force-test a creative Meta/TikTok won't serve, put it alone in its own campaign with a fairly high budget.

Founder, who built the app for himself after reading David Goggins, posted on TikTok and Instagram about going through the challenges, with little money at the start. Suggests AI-generated talking-head videos of your own face for cheap testing.