Daily active users (DAU) is the count of unique users who engage with a product within a single calendar day. "Engage" typically means any meaningful interaction: opening the app, completing an action, triggering an event. The precise definition varies by product and should be documented explicitly, because how you define "active" determines what DAU actually measures.
DAU is one of the most widely reported product health metrics, but its value in experimentation is more nuanced than its popularity suggests. As a guardrail or secondary metric in an A/B test, DAU can signal whether a change is driving users away or pulling them back. As a success metric, it's often too blunt to capture the effect of a single feature change on a product with millions of users.
How is DAU used in experimentation?
In experiment analysis, DAU matters at the per-group level. You're comparing whether the treatment group has more or fewer active users than the control group on a given day. A drop in DAU in the treatment group signals that the change may be causing users to disengage.
DAU works best as a guardrail metric or a secondary metric rather than as a primary success metric for most experiments. The reason is sensitivity. A product-wide DAU metric is influenced by everything happening in the product simultaneously: seasonality, marketing campaigns, other concurrent experiments. The signal from one feature change is typically small relative to that noise. Confidence's analysis infrastructure supports metric definitions scoped to specific product surfaces, which lets you measure engagement at the feature level rather than diluting the signal across the entire product.
Where DAU becomes more useful is in aggregate: tracking it as a longitudinal guardrail across many experiments to detect whether the cumulative effect of shipped changes is growing or shrinking the active user base over time.
How does DAU differ from WAU and MAU?
DAU captures the most granular view of engagement. It's volatile: weekdays differ from weekends, holidays create dips, and a single push notification campaign can spike DAU temporarily. This volatility makes DAU responsive to short-term changes but noisy for long-term trends.
WAU (weekly active users) smooths out daily fluctuations by counting unique users over a 7-day window. MAU (monthly active users) smooths further across 30 days. Each level of aggregation trades sensitivity for stability. In experiments, DAU is the most likely of the three to detect an effect within a short experiment window, but it's also the most likely to produce spurious day-to-day fluctuations that complicate interpretation.
What counts as "active"?
This is the most important decision in defining DAU, and it's the one most often left implicit. A user who opens the app for one second and closes it is "active" by some definitions and not by others. A user who receives a push notification and taps it has engaged, but the engagement was prompted rather than organic.
Spotify tracks engagement metrics with specific event definitions: a listening session counts differently from opening the app, which counts differently from interacting with a feature. The right definition of "active" for your DAU metric depends on what question you're trying to answer. Document the definition, and keep it consistent across experiments so results are comparable.