Active Users Analysis

Active Users Analysis measures the scale of active users of your product. Its core metrics are DAU (daily active users) / WAU (weekly active users) / MAU (monthly active users), making it the foundational model for assessing product health and user stickiness. It corresponds to the backend AnalyticsService.ActiveUsers.

1. What Questions It Answers

  • How many active users are there today (DAU)? What is the trend?
  • How do WAU and MAU compare to DAU? (reflecting stickiness)
  • Is DAU rising or falling over the last 30 days?

2. Key Metrics

MetricMeaning
DAUDistinct active users on the day
WAUWeekly actives: distinct active users within a rolling 7-day window
MAUMonthly actives: distinct active users within a rolling 30-day window
WAU/DAUReflects weekly revisit frequency (the closer to 1, the more users come every day)
MAU/DAUReflects how diluted the monthly active base is relative to the current day

3. Backend API

gRPC: AnalyticsService.ActiveUsers

FieldTypeDescription
time_rangeTimeRangeAnalysis time range
granularityAnalyticsGranularityTime granularity (DAU is bucketed by granularity)
app_iduint32 (optional)Filter by app (0 = all tenants)

Response ActiveUsersResponse: points[] (each ActiveUsersPoint contains timestamp / dau / wau / mau), latest_dau (DAU of the latest day).

HTTP (admin proxy)

POST /admin/v1/analytics/active-users
Content-Type: application/json

{
  "timeRange": { "startMs": 1718169600000, "endMs": 1718774399000 },
  "granularity": "DAY"
}

AnalyticsGranularity values: UNSPECIFIED(0) server auto-selects / HOUR(1) / DAY(2) / WEEK(3) / MONTH(4).

4. Implementation

  • DAU: distinct count of events_fact.user_id per day (count(DISTINCT user_id)).
  • WAU / MAU (daily level): returns accurate values based on HLL rolling windows — the pre-aggregation table stores the daily HLL_UNION(HLL_HASH(user_id)) state, and WAU/MAU are obtained by merging the daily uv states within the rolling window with HLL_UNION.

⚠️ Hourly granularity limitation: since there are no hourly uv states, at HOUR granularity WAU/MAU degrade to equaling DAU (provided as a lower bound only). Daily and coarser granularities return accurate values. See Appendix · Known Gaps.

5. Typical Scenarios

Daily operations dashboard

Query DAU/WAU/MAU for the last 30 days at DAY granularity, plot them as a trend chart, and monitor fluctuations in the active user scale.

Stickiness assessment

Compute WAU / DAU: a ratio close to 1 means users come almost every day (high stickiness, e.g. tool-type products); a larger ratio means users are spread across different days of the week (e.g. content-type products).

6. Notes and Caveats

  • Deduplication basis: DAU is deduplicated by user_id; anonymous users (no userId) are not counted in DAU — bind users with identify first.
  • Rolling windows: WAU means "the last 7 days" rather than "a calendar week"; MAU means "the last 30 days" rather than "a calendar month".
  • HLL approximation: WAU/MAU are based on HLL, so results are approximate (error typically < 2%) — suitable for trend monitoring rather than exact reconciliation.
  • Empty data: if DAU is 0, first confirm events have been stored (see Getting Started · Current State of Data Storage).

7. Related Documents