Objective: Our objective was to assess low-cost and feasible feedback alternatives and compare them to Lambert’s OQ feedback system. We also studied patient, therapist, and process characteristics that could moderate the effect of feedback on outcome, session attendance, and alliance. Method: A total of 547 patients, 75% female, average age 41 (SD = 13), 95% Latino, treated in an outpatient individual psychotherapy setting in Chile were randomly assigned to five feedback conditions: no feedback, feedback on symptomatology, feedback on the alliance, feedback on both symptomatology and alliance, and Lambert’s OQ progress feedback report. The measures included the Outcome Questionnaire and the Working Alliance Inventory. We also had follow-up interviews with therapists. Results: We found through multilevel modeling that feedback had no effect on outcome, session attendance, and alliance. Contrary to previous findings, these results were maintained even when focusing only on patients “not-on-track.” However, patients’ former psychiatric hospitalization history and baseline severity, combined with lack of progress, significantly moderated the impact of feedback. For this more dysfunctional population, “positive feedback” (i.e., low symptomatology) to therapists had a positive impact on therapy outcome, while “negative feedback” (i.e., high symptomatology) had a negative impact. Conclusions: Providing feedback to therapists without offering them tools to improve treatment may be ineffective and even be detrimental. This may be especially the case for patients who suffer more severe mental health issues and whose therapists receive mostly discouraging news as feedback.