newsfilter.io
Lecture

Dot Plots: How to Actually See What Your Users Are Doing

  • Founders frequently rely on aggregate metrics (e.g., DAUs, MAUs) which obscure individual user behaviors, masking issues with product fit even when total numbers grow.
  • The speaker proposes using "dot plots" to visualize individual user activity over time, treating rows as users and columns as time periods (typically days).
  • Dots within the grid represent specific, value-driving events (e.g., listening to a song, sharing a photo, processing an invoice) rather than generic actions like opening an app.
  • A distinct symbol (e.g., a ring) can mark a user's first day of activity (onboarding) to better visualize retention patterns.
  • Dot plots allow for the identification of usage patterns invisible in aggregate data, such as:
    • Distinct cohorts, such as "weekday workers" versus "weekend-only" users.
    • Immediate churn, where users activate once and never return.
    • Feature correlation, such as a user who joins a public playlist subsequently showing high consecutive usage.
  • The methodology scales from startups to massive enterprises:
    • Early-stage founders can review every single user on one screen.
    • Large-scale operations (e.g., Google Photos with >1 billion users) use sampling, printing specific dot plot segments for targeted teams (e.g., "iOS users in France" or "US users earning >$80k").
  • The technique originated from a fraud detection strategy used by PayPal at the time of its launch, where analysts visually scanned transaction graphs to identify anomalies.
  • In B2B contexts, dot plots reveal usage gaps that threaten contract renewal, such as:
    • A case study where a company purchased 10 seats but only 3 were activated.
    • Sporadic usage patterns (max 2 days/week) that signaled a lack of value before a key champion left the client organization.
  • Critical implementation errors include:
    • Selecting "vanity" events that feel good but do not indicate real value (e.g., app opens).
    • Using time intervals that are too broad (e.g., weeks instead of days), which hinders the detection of specific behavioral rhythms.
  • The speaker recommends dot plots as the primary dashboard for products with fewer than hundreds of users.
  • Dot plots should be used in conjunction with cohort retention curves:
    • Retention curves determine if user groups stick over time.
    • Dot plots explain how those users interact and provide the context to ask the right questions for product iteration.
  • Modern AI coding tools can generate these visualizations in approximately 10 minutes by parsing raw logs into a 2D grid format.