In the second stage of unification, the candidate groups are divided into client groups.

The client groups are created by repetitive selection of the best record (center) from the remaining records of the candidate group. Then all records (slave) that are similar are added to this center.

The rating of quality is defined by the center selection rule, and similarity by the matching rule. The number of such center selections, and thus the number of client groups in one candidate group, is limited. Potential records from a candidate group that have not been assigned to a center are called renegades.

Each client group is identified by a number called a Client ID.