Skip to main content

Average Daily Users

Distinct users on a typical day over a reporting window.

How many distinct users signed in to CanadaLogin on a typical day over a selected reporting window: the mean of the daily unique-user counts.

It is a typical day, not reach. An average of 2,000 daily users over four weeks does not mean 2,000 people used CanadaLogin in those four weeks; the same person on ten different days counts on all ten. For how many different people used it, read a cumulative column (see Cumulative Unique Users). There is no unique-user count for an arbitrary window such as “the last 28 days”.

Calculation

Average unique_users over the days in the window that have a row. Return the number of days present beside it, and quote the average only when that equals the number of days in the window. A missing day is a gap in the data, not a day with no users: dividing by the length of the window would count it as zero and pull the average down.

SELECT
    AVG(CAST(unique_users AS DOUBLE)) AS average_daily_users,
    COUNT(DISTINCT CAST(from_date AS DATE)) AS days_present,
    -- the window's length, inclusive; quote the average only if they match
    date_diff('day', DATE '2026-09-03', DATE '2026-09-30') + 1 AS days_in_window
FROM ibm_verify_events_transformed.events_auth_total_logins
-- set the reporting window; see the timespans note in cookbook/notes.qmd
WHERE CAST(from_date AS DATE) BETWEEN DATE '2026-09-03' AND DATE '2026-09-30';
tbl(con, in_schema("ibm_verify_events_transformed", "events_auth_total_logins")) |>
  # set the reporting window; see the timespans note in cookbook/notes.qmd
  filter(
    as.Date(from_date) >= as.Date("2026-09-03"),
    as.Date(from_date) <= as.Date("2026-09-30")
  ) |>
  summarise(
    average_daily_users = mean(as.double(unique_users), na.rm = TRUE),
    days_present = n_distinct(as.Date(from_date)),
    # the window's length, inclusive; quote the average only if they match
    days_in_window = !!(as.integer(as.Date("2026-09-30") - as.Date("2026-09-03")) + 1L)
  )

Interpretation notes

Say “on a typical day” wherever it appears. Readers take “average users” for how many people use the service, which is a much larger number.

Comparing two windows. For a change against the previous window, use windows of the same length and check both are complete. Weekends run lower than weekdays, so a window should cover whole weeks: 28 days is four of each weekday.

The rate behind other metrics. Summing the same daily counts instead of averaging them gives user-days, the denominator of Calls per Active User.

Alternative names

  • daily active users
  • DAU
  • average daily active users
  • typical daily users