This document describes a collaborative filtering approach using co-clustering with augmented data matrices (CCAM). CCAM extends a co-clustering algorithm based on information theory to simultaneously cluster users, items, and additional data (e.g. user profiles, item features). The authors apply CCAM to collaborative filtering by using the co-clusters as prototypes for predicting user ratings. They tune CCAM's parameters on a dataset from an online advertising site and compare its mean absolute error to other collaborative filtering methods. CCAM outperforms k-means clustering, k-nearest neighbors, and information-theoretic co-clustering on this task.