Evaluating if Your Search Interface Produces Enough Data for ART

The team behind Coveo™ Machine Learning establishes minimal requirements for the Automatic Relevance Tuning (ART) feature to deliver the best results [see Automatic Relevance Tuning (ART) Feature]. A search interface candidate, for a specific language, must have 5,000 visits over the past three months that contained at least a manual query followed by a click.

ART works with less data, however improvements on metrics such as the click-through and the average click rank are going to be less significant (see Note: Number of events).

ART requires real data from a specific community to ensure ART recommendations respond to real user intentions. It is thus strongly NOT recommended to provide false generated data.

You can use the Analytics section of the administration console navigation menu to browse user visits, and thus evaluate if a search interface respects the ART feature minimal requirements (see Required Privileges).

The following procedure takes into account that you are familiar with global dimension filters and Visit Browser features (see Reviewing User Visits With the Visit Browser and Adding Global Dimension Filters).