How To Better Understand Your Users

The speaker argues that founders commonly make the mistake of relying on aggregate user metrics (DAUs, MAUs) instead of understanding how individual users interact with their products. Aggregate graphs tend to show upward trends even when users aren't genuinely engaged, masking critical behavioral insights.

The speaker introduces the dot plot as a solution: a two-dimensional grid where rows represent individual users and columns represent time periods (typically days). A dot is placed in each cell when a user performs a key value-creating action (like listening to a song in Spotify or sharing a photo). The first day a user onboards can be marked with a special symbol (like a ring around the dot) to track initial engagement.

When visualized as a dot plot, patterns emerge that aggregate metrics completely obscure. The speaker's Spotify example reveals that some users engage on weekdays while others use the product only on weekends—an insight invisible in DAU graphs. Similarly, the visualization quickly exposes one-time users who never return, signaling onboarding problems.

The tool can be made more sophisticated by using different symbols to represent different features or events, and by encoding user attributes (device type, location, demographics) through color, shading, or annotations. Rows can be sorted by these attributes to compare cohorts.

For B2B products, dot plots prove equally valuable. The speaker describes a YC company with an $80,000 customer contract that churned after the product champion left the company. A dot plot would have immediately revealed that only 3 of 10 purchased seats activated and usage was sparse and inconsistent—red flags about contract health that aggregate renewal data wouldn't catch.

The speaker warns against two common mistakes: charting low-value events (like app opens) instead of genuine value-creation events, and using time periods that are too wide (weeks instead of days), which obscures actual behavior patterns.

Dot plots scale from small user bases (fitting on one screen) to billions of users through sampling—the speaker describes practices at Google Photos where teams would print dozens of dot plots representing different user segments (iOS users in France, US web users earning $80K+) and analyze patterns manually.

The speaker positions dot plots and cohort retention curves as complementary tools: retention curves show in aggregate whether user cohorts stick around, while dot plots reveal how users actually behave within the product, directing founders toward better questions and features.