See How Customers Actually Use Your Product
In a Nutshell
Dot plots reveal individual user behavior over time by placing a dot for each meaningful usage event per user per day, exposing patterns like weekday-only usage, feature-driven retention, and low activation that aggregate metrics hide. The key is choosing the right "value event" (e.g., listening to a song, processing an invoice) rather than superficial actions like app opens. Pairing dot plots with cohort retention curves gives founders the clearest signal on whether users are getting real value and which features drive sustained engagement.
These notes were generated by AI and may contain inaccuracies.
One of the biggest mistakes founders make is relying on aggregate user metrics instead of understanding how any individual users use their product. Cohort retention curves are the best tool to figure out if people keep using your product. However, these curves do not reveal how users are actually using the product, which features they interact with, the frequency of use, or the pacing of usage. Most founders ignore this information, but it is the most important signal to determine if a product has been built that people want.
Founders need to look at what individual users are doing. Even with 10 or 20 users, tailing logs and watching every event is challenging. Aggregate data such as DAUs or MAUs lump all users together, making it impossible to understand individual behavior. With growth, these graphs tend to trend up and to the right even if users are not actually enjoying the product.
A dot plot is a tool that allows understanding of individual users while also providing a big picture view of product performance. The visualization is a two-dimensional grid similar to a spreadsheet with rows and columns. Each row represents one individual user. Every column represents a time period, typically days depending on the nature of the product.
The idea is to put dots in each cell representing an event that users perform that indicates value in the product. Examples include sharing a photo in a photo app, listening to a song in a music app, or processing an invoice in a B2B invoice processing product. A dot is placed for each day that each user uses the product.
For a Spotify music streaming app, the event of listening to a song is charted. Anytime a user listens to a song during a day, a dot is placed. A user's first day of using the product can be marked with a ring around the dot to indicate onboarding.
The visualization creates a high-density view of individual users and their usage over time. It reveals patterns that would not be visible through aggregate charts or individual user logs.
Looking at the Spotify example, several patterns emerge. Some users use the product on weekdays while others only use it on weekends. This observation could inform product redesign, user targeting decisions, or understanding which users are most valuable. Without a dot plot, these behavioral differences would remain invisible.
Dot plots also provide a measure of retention. Users who try the app on one day and never return indicate potential problems with onboarding or other aspects of the product.
At Bump, different symbols were used in cells to distinguish between actions such as sharing contact information versus sharing a photo. This provides greater granularity and allows deeper analysis as needed.
The dot plot concept is similar to GitHub contribution graphs, which wrap days around per week. Beyond tracking actions, dot plots can track user state such as device type (iPhone, Android, web) or geographic location (United States or another country). Demographic information such as income level can also be encoded using symbols or cell shading.
Rows can be sorted based on any attributes. For example, viewing only iOS users or only users whose first day using the app was a specific Monday becomes possible.
This visualization approach was used at PayPal when Max Levchin and the team faced a significant fraud problem. They built a graph of all transactions and had humans sit and stare at the screens. The observers identified anomalous patterns that indicated fraud without knowing exactly what they were looking for beforehand. They would then investigate those specific patterns. Dot plots work similarly, allowing patterns to be spotted visually before deeper investigation.
A DAU graph for the same users might show numbers such as 2, 3, 2, 2, 2, 2, 1, 0, 1 across consecutive days. This reveals very little beyond the fact that the product is not growing and has some users. The dot plot provides a much richer understanding of user behavior and potentially their lives, such as inferring that weekday users may be listening at the office.
Dots can represent different features. In the Spotify example, an S could indicate search usage and a P could indicate joining a public playlist. Patterns may reveal that specific features drive desired behaviors. A user who joined a public playlist and then showed many consecutive days of usage might indicate that the playlist feature is causal to high engagement.
For early-stage founders with few users, every single user and every day of usage fits on one screen. The approach scales to thousands, millions, or billions of users through sampling. At Google Photos with over a billion users, dozens of dot plot printouts were created for different user base samples. Team members would receive specific segments such as iOS users in France or web users in the United States earning more than $80,000 per year to analyze.
Dot plots are useful for B2B products that sell seats to businesses. In one YC batch company, a name-brand customer signed an $80,000 annual contract for 10 seats. After onboarding, only three seats ever activated. Those three users showed sporadic usage, never exceeding two days per week. The champion who purchased the product left the company, and the new person churned the contract. The dot plot would have revealed the contract was in jeopardy through the low activation and usage patterns.
The primary mistake is charting the wrong event. Founders may choose easy events like opening the app or signing in because they produce many dots, but these do not measure whether users receive real value. The recommended approach is to select events that represent actual value creation for the user.
Another mistake is choosing time periods that are too wide, such as weeks instead of days. Day-level or even sub-day granularity makes it easier to understand actual usage patterns.
Until reaching hundreds of users, the dot plot could serve as the only dashboard. Dot plots are fundamentally a logs visualization tool requiring no fancy computations, only parsing logs into a 2D grid. Modern AI coding tools can create this visualization in approximately 10 minutes.
Dot plots work best alongside cohort retention curves. Cohort retention curves show whether acquired user groups stick with the product over time in aggregate. Dot plots reveal how those users actually use the product. Together they provide the information needed to ask the right questions of users, build the right features, and fix broken product elements that aggregate metrics would never reveal. These two tools are among the most important for understanding users.
Keep Y Combinator in your library
Save the videos and channels worth coming back to, and find them again in one place.
