Lecture 6 - Growth (Alex Schultz)

If your cohort retention curve doesn’t asymptote (flatten) and instead slopes to zero, you don’t have durable product-market fit—so virality, ads, and “growth hacking” will only amplify churn.

By measuring what percent of users are active on day 31, 32, 33, etc. after signup, you can infer the retention shape with relatively small samples (e.g., ~10k users) and make surprisingly accurate value predictions.

Retention targets differ by vertical; use back-of-the-envelope market sizing (e.g., internet users vs. actives) and comparable companies to triangulate what “successful” terminal retention must look like.

Once teams scale, control becomes influence; a single metric (MAU, sends, nights booked, GMV) prevents local optimization and keeps product, marketing, and engineering aligned.

Activation isn’t abstract: for Facebook it’s “see your friends” (operationalized as ‘10 friends in 14 days’), and for marketplaces it’s finding the unique item or getting paid the first time.

Power users rarely leave due to “too many notifications”; marginal users often churn because they never reached the core value (e.g., low friend count), so growth should target their activation and re-engagement.

Facebook’s community translation system enabled rapid, ongoing language expansion (104 languages, many community-driven) and avoided optimizing only for “today’s” biggest languages as the world shifted (e.g., Hindi growth).

Hotmail won via high frequency and conversion; PayPal won via extremely high intent (someone wants money) plus incentives—yet none of it matters if retention behind the loop is weak.

Keyword research prevents optimizing for low-demand terms; external links drive ranking, and internal architecture (e.g., adding a directory) can unlock orders-of-magnitude traffic gains.

Email/SMS/push all fail if you get spam-foldered or users opt out; the highest leverage is timely, user-specific notifications (especially for low-engagement users), not one-size-fits-all blasts.