Not every identifier in your source data will connect immediately in the Identity Graph. Coverage (sometimes referred to as “match rate”) reflects the proportion of User IDs or events that are successfully linked to others, either at the Household or Person level.
Coverage depends on two key factors:
- Whether multiple identifiers are available for a user or household in your source data.
- Whether sufficient behavioral patterns exist for the algorithms to confirm a match.
Types of Coverage Metrics
1. User ID–Based Coverage
The percentage of all User IDs in your source data that have been linked to at least one other ID. In other words, the share of identifiers that are connected to a Household ID or Person ID cluster.
2. Event-Based Coverage
The percentage of all events (e.g., impressions, bids, log entries) that carry a User ID which is connected to at least one other ID.
Because some IDs are very active (appearing in many events) while others are rare (seen only once), User ID coverage and Event coverage can differ significantly.
3. By ID Type
If multiple types of identifiers are present (cookies, MAIDs, TV IDs, HEMs, etc.), coverage can be reported separately by type to give more insight into data quality and match performance.
Look-back vs. Real-time Coverage
Coverage can also be viewed in two “windows”:
-
Look-Back Coverage
Calculated using the most recent day of source data at the time the graph was built.- Best used for analytics scenarios such as reach & frequency, or attribution.
-
Real-Time Coverage
Calculated on the following day’s data, after the graph was computed.- Best used for activation and targeting.
- Real-time coverage is typically slightly lower, since newly observed User IDs may not yet be incorporated into the graph.
Household vs. Person Coverage
- Household coverage: Broadest level of resolution. Most identifiers will link into households, including shared devices that cannot be tied to a single user.
- Person coverage: More granular and precise. Coverage is slightly lower, since not all identifiers provide enough evidence to resolve to a specific person.
📘 Key takeaway: Coverage rates are a function of your input data and the natural dynamics of identifiers. Household-level coverage provides a strong foundation for measurement, while Person-level coverage adds more precision where signals allow.