a16z Podcast: The Basics of Growth 2 -- Engagement & Retention

Episode Summary

Executive Summary: This episode examines how growth shifts from acquisition to engagement and retention as products mature. The hosts explain cohort analysis, churn/resurrection accounting, and how metrics like DAU/MAU, L28, and usage curves reveal product quality, power users, and network effects. They emphasize that the right metric depends on the business model and cadence of use.

Main Topics: From acquisition to engagement and retention (Priority: 5/5): As companies scale, the growth focus moves from adding new users to improving engagement, retention, and resurrection of lapsed users. The leverage changes by company stage and market saturation. Cohort analysis and user quality (Priority: 5/5): Cohorts are used to compare user batches over time, track decay or plateau, and identify whether later cohorts are better or worse than earlier ones. This helps isolate product changes and market effects. Network effects revealed through metrics (Priority: 5/5): True network effects show up in cohort curves improving over time, not just in claims. More users should make the product more valuable, and this can vary by market density or team size. Engagement ladders and onboarding (Priority: 4/5): Products often require users to progress from an initial valuable action to a higher-frequency, higher-value behavior. Onboarding, content, education, prompts, and incentives can move users up that ladder. Choosing the right engagement and retention metrics (Priority: 5/5): DAU/MAU, frequency, L28, L7, time spent, and churn each measure different things. The hosts stress matching metrics to the product's natural cadence rather than forcing daily-use expectations onto episodic products. Engagement, retention, and monetization are distinct (Priority: 4/5): Retention is not the same as engagement: some products are low-frequency but highly retained, while others are highly engaging but naturally episodic. Monetization strategy should align with the actual usage pattern. Metrics as a defense against misleading growth (Priority: 4/5): Growth can be gamed with paid acquisition or notifications, but engagement metrics are harder to fake. The best companies use deep metric analysis to understand behavior and improve the product iteratively.

Key Arguments: Early-stage companies prioritize acquisition, but as they scale, churn, retention, and resurrection become the real growth levers. Cohort curves are a practical way to evaluate whether a product is improving, degrading, or benefiting from network effects. If later cohorts perform better than earlier cohorts, the product is likely becoming more valuable over time rather than simply benefiting from one-time hype. Segmenting cohorts by geography, team size, or market can reveal natural A/B tests and show how density affects engagement. Aha moments are often not a single event; they can be engineered through onboarding, content, setup, and repeated exposure to product value. DAU/MAU is useful for high-frequency products like advertising-driven social platforms, but it can mislead for episodic products like travel. L28 and related histograms help identify hardcore users and power-user clusters by showing how often users return within a 28-day window. Engagement metrics are hard to game compared with growth metrics because casual notification spam may inflate actives without improving true usage. Network effects should be demonstrated through data, such as declining churn, rising conversion, and improved per-market performance, not asserted rhetorically. The right retention benchmark depends on product cadence; a twice-a-year travel app should not be judged like a daily messaging app.

Data Points: DAU/MAU: 60%+ - Facebook historically had daily actives above 60% of monthly actives, illustrating a high-frequency product. Pinterest penetration: Most women in America have downloaded the Pinterest app - Used to illustrate market saturation in the U.S. and a shift toward retention and re-engagement. OpenTable median usage: Twice a year - Shows an episodic product where low frequency is normal and does not imply poor retention. L28 window: 28 days - Histogram-based metric used to show how many days a user visits within a typical month and reduce seasonality. L7 window: 7 days - Shorter-window frequency measure referenced as a related retention/usage metric. Network effect example: 10x more restaurants - Illustrates how more inventory in a marketplace like OpenTable can increase reservations per diner. Rideshare example: 15 minutes vs 10 minutes - Used to show diminishing marginal returns in network effects; reducing wait times from 15 to 10 minutes matters more than from 5 to 2 minutes. Market density example: 100 restaurants in Des Moines vs 100 in Manhattan - Shows why raw counts must be normalized by market size when analyzing network effects. Product behavior example: 20 times a day - San Francisco weather checks were mentioned as an extreme high-frequency use case. Notification effect: Can decrease DAU/MAU - Sending more email/push notifications can raise monthly actives via casual users while not increasing daily actives proportionally.

Pivotal Quotes: "show me the data, cohort. It's like, show me the money is now show us the cohort." — Jeff Jordan / Andrew Chen: On proving network effects and product value through cohort analysis rather than claims. "Growth is good. Growth and engagement is really, really good." — Jeff Jordan / Andrew Chen: Closing takeaway on how growth alone is insufficient without sustained engagement. "It's not just an accident. You have to sort of architect it, not just expect like serendipity to fall into place." — Jeff Jordan / Andrew Chen: On designing onboarding and engagement pathways instead of relying on luck.

Implications: For founders and investors, the key is to match metrics to product cadence, prove network effects with cohorts, and focus on deep engagement once acquisition matures. Misreading DAU/MAU or churn can lead to bad strategy.