Data retention

Coveo typically retains usage analytics data for an extended period to allow you to produce reports showing long-term trends and comparing similar periods over months, quarters, and even years.

However, UA data is retained for a minimum of 1 year to a maximum of 3 years to maintain performances and keep the database size manageable. As shown in the following table, the data retention period is shorter for some specific dimensions (see Concepts).

Typical period Data description

3 years

Standard and custom events with associated dimensions

1 year

customData and the Advanced Query Expression dimensions

  • The data retention period starts for a specific dataset when the corresponding usage analytics event occurs.

  • Coveo can permanently delete usage analytics data when the data retention period ends.


    You started gathering usage analytics data on May 4th, 2021. With a three-year data retention period, data recorded on May 4th, 2021, can be permanently deleted starting on May 5th, 2024.

  • Your subscription agreement with Coveo may specify other usage analytics data retention periods.

About the customData object

The customData object contains a set of metadata (key-value pairs) in JSON format (see Concepts). The customData object has a shorter retention period because in the UA database, its content is generally replicated in individual custom dimensions and the original JSON raw data increases the database size without adding value.

Picture the UA database as a table; each row corresponds to an event and each column contains the value of one dimension. The customData object, however, is special as it’s stored in one column as a whole in JSON format.

When an administrator maps one of the customData metadata to a custom dimension (see Manage dimensions on custom metadata), a column is created in the database for the custom dimension and is populated with the corresponding values from past and future events, therefore duplicating the data. The three-year data retention period applies to custom dimensions.

Consequently, when all customData metadata are mapped to a custom dimensions within a year from the time they start to be collected, the customData one-year data retention period has no data availability impact, and can even contribute to improved performance.

About the Advanced Query Expression dimension

The Advanced Query Expression dimension has a shorter retention period because it impacts database size and performance, and is typically useful mainly for debugging purposes and only shortly after event occurrences.

The Advanced Query Expression dimension contains the part of the query that’s hidden and typically includes filters for the search interface scope and facet selection (see Search dimensions). This expression can often be long, increase the database size, and rarely used in filters. However, when it is used in filters, parsing its content may affect performances.