ibm_verify_events_transformed.events_app_login_counts
Daily SSO event counts for each CanadaLogin-integrated application.
Each row is one application on one day. It counts what IBM Verify calls the sso event type, raised when someone who is already signed in reaches an application.
events_auth_total_logins counts a different event type (authentication). Never add the two tables’ figures together or show them side by side.
Keywords: SSO events, application usage, sign-ins, logins, applications, unique users, adoption, daily counts
Provenance
Built from application_usage by compute_app_login_counts() in the daily metrics job.
- Updated: Daily, at 04:00 ET, for the previous ET day
- Retention: While CanadaLogin exists
- Steward: EDCP Data and Research Team
- Contact: Yinlin Zhang
- Location:
s3://<transformed-bucket>/ibm_verify_metrics/events_app_login_counts/
Fields
One row per application_id per day.
| Field Name | Type | Description |
|---|---|---|
| application_id | string | IBM Verify’s numeric ID for the application. Stable when the application is renamed. Join to rp.alias on this. "UNKNOWN" marks events with no application |
| application_name | string | The latest name seen for the application that day (for example, “IRIS - PRC - Prod”). It changes if the application is renamed, so do not group or join on it. For a label, use the latest name in events_application_mapping |
| from_date | string | The reporting day in ET, for example “2026-09-15”. Cast before comparing to a date |
| to_date | string | Same as from_date |
| total_logins | bigint | SSO events for the application that day. Counts events, not people |
| unique_users | bigint | Distinct users with an SSO event for the application that day, successful or not |
| failed_logins | bigint | Failed SSO events that day |
| successful_logins | bigint | Successful SSO events that day |
| mtd_unique_users | bigint | Distinct users of the application month-to-date. Resets on the 1st |
| ytd_unique_users | bigint | Distinct users of the application year-to-date. Resets on January 1 |
| fetched_at | string | When the job wrote the row, in ISO 8601 UTC |
| date | string | Partition key. Same as from_date |
Notes
total_logins, unique_users, failed_logins, and successful_logins are daily values. The mtd_ and ytd_ columns are cumulative: take the latest value in the window, never sum them.
One row per ID. An application keeps its ID when it is renamed, so group on application_id, not application_name. If a service has more than one application, summing unique_users across them double-counts people who used both.
Join to relying parties after, not before, aggregating. Several applications can map to one service. See Joining through the lookup.
Failed events rarely have a user. Almost all failed SSO events have the user ID "UNKNOWN", which counts as one user.
Not inflated by second factors. Reaching an application uses the session the person already holds, so one handoff is one SSO event. total_logins still counts events, because one person can reach an application many times.