Contracting out public services to private firms has ambiguous effects when quality is imperfectly observable. Using a randomized experiment over a national sample in France, we compare the efficiency of the public employment service (PES) vs. private providers in delivering very similar job-search intensive counseling. The impact of each program is assessed with respect to the standard, low intensity track offered by the PES to the unemployed. We find that job-search assistance increases exit rates to employment by 15 to 35%. But the impact of the public program is about twice as large as compared to the private program, at least during the 6 first months after random assignment. We argue that the observed contract structure with the private providers has not overcome the underlying agency problem. We find no evidence of cream-skimming: rather, it seems that profit maximizing private providers have found it optimal to enroll as many job-seekers as they could, but to make minimum effort on the placement of some of them.