Results from an earlier clinical trial comparing narrative therapy with cognitive–behavioural therapy (Lopes et al., 2013) suggested that narrative therapy is efficacious for depression. However, there were significant differences in symptom reduction on the Beck Depression Inventory‐II, favouring cognitive–behavioural therapy, if dropouts were included in the analysis, suggesting that time to recovery or improvement would differ in both treatments. Contrarily, results showed that treatment assignment was not a predictor for differential effect. Using a survival analytic approach, it was found that four sessions were necessary for 50% improvement and 16 sessions for 50% recovery. Additionally, depressive symptoms changes occurred significantly faster than interpersonal changes, again regardless of treatment assignment. These results support previous findings of the dose–response literature and of the phase model of change, with the advantage of being specific to psychotherapy with depressive clients. Copyright © 2013 John Wiley & Sons, Ltd. Key Practitioner Message For 50% of clients with major depressive disorder, it takes four sessions to improve and 16 sessions to recover, regardless of whether they were treated with narrative therapy or cognitive–behavioural therapy. For those clients who recover, they do so by session 11.Clients change depressive symptoms more consistently and much faster than they change interpersonal problems. For clients who will not recover during brief interventions and especially for clients who present strong interpersonal problems at onset, long‐term treatment plans should be considered.More emphasis should be laid on symptomatic relief in the early stages of treatment and on interpersonal issues at later stages.