Skip to main content

Authentication Success Rate

Share of authentication attempts that succeeded, across CanadaLogin.

The proportion of authentication attempts that succeeded, across all of CanadaLogin. A failed attempt can be a wrong password or a failed second factor.

This metric is useful for monitoring overall authentication health. A sustained drop can indicate a service problem, a misconfigured application, or elevated attack traffic inflating the failure count.

Calculation

Divide the sum of successful_logins by the sum of all attempts (successful_logins + failed_logins) over the selected reporting period.

To track the metric over time, bin the same calculation by week. Truncating from_date to the week groups each day into its Monday-anchored bin, so the result is one success rate per Monday-to-Sunday week.

Overall:

SELECT
    CAST(SUM(successful_logins) AS DOUBLE)
        / NULLIF(SUM(successful_logins + failed_logins), 0) AS authentication_success_rate
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 '2026-01-01' AND DATE '2026-05-28';

By week:

SELECT
    date_trunc('week', CAST(from_date AS DATE)) AS week_start,
    CAST(SUM(successful_logins) AS DOUBLE)
        / NULLIF(SUM(successful_logins + failed_logins), 0) AS authentication_success_rate
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 '2026-01-01' AND DATE '2026-05-28'
-- date_trunc('week', ...) anchors each bin to its Monday
GROUP BY date_trunc('week', CAST(from_date AS DATE))
ORDER BY week_start;

Overall:

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("2026-01-01"),
    as.Date(from_date) <= as.Date("2026-05-28")
  ) |>
  summarise(
    authentication_success_rate =
      sum(as.double(successful_logins), na.rm = TRUE) /
        sum(successful_logins + failed_logins, na.rm = TRUE)
  )

By week:

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("2026-01-01"),
    as.Date(from_date) <= as.Date("2026-05-28")
  ) |>
  # floor_date(..., "week") translates to Athena's date_trunc('week', ...),
  # which anchors each bin to its Monday
  mutate(week_start = floor_date(as.Date(from_date), "week")) |>
  group_by(week_start) |>
  summarise(
    authentication_success_rate =
      sum(as.double(successful_logins), na.rm = TRUE) /
        sum(successful_logins + failed_logins, na.rm = TRUE)
  ) |>
  arrange(week_start)

Interpretation notes

The rate is a proportion between 0 and 1. Read it alongside authentication event volume: a low success rate on very low volume may just be noise, while the same rate on high volume is a real signal. In the weekly view, the first and last weeks of the reporting window are usually partial (the window rarely starts or ends on a Monday), so treat boundary weeks with caution when comparing trends.

This rate is computed over authentication methods, not sessions. When signing in to CanadaLogin, IBM Verify records each method as its own event, so a mistyped password is a failed event even when the user then succeeds and completes the sign-in (see the events_auth_total_logins caveat). The figure measures how often individual authentication steps succeed, not how often a sign-in ultimately completes. The per-application rate counts a different IBM Verify event type, the handoff to a relying party with no factors entered, so the two rates are not comparable.

Alternative names

  • sign-in success rate
  • overall authentication success rate
  • successful authentication proportion