Collaborative Filtering with Implicit Feedbacks (e.g., browsing or clicking records), named as CF-IF, is demonstrated to be an effective way in recommender systems. Existing works of CF-IF can be mainly classified into two categories, i.e., point-wise regression based and pair-wise ranking based, where the latter one relaxes assumption and usually obtains better performance in empirical studies. In real applications, implicit feedback is often very sparse, causing CF-IF based methods to degrade significantly in recommendation performance. In this case, side information (e.g., item content) is usually introduced and utilized to address the data sparsity problem. Nevertheless, the latent feature representation learned from side information by topic model may not be very effective when the data is too sparse. To address this problem, we propose collaborative deep ranking (CDR), a hybrid pair-wise approach with implicit feedback, which leverages deep feature representation of item content into Bayesian framework of pair-wise ranking model in this paper. The experimental analysis on a real-world dataset shows CDR outperforms three state-of-art methods in terms of recall metric under different sparsity level.