Background: Treatment of depressive disorders in old age is hindered by several barriers. Most common are time pressure in primary care and latency for specialized therapeutic care. To improve treatment, the collaborative care approach GermanIMPACT was evaluated in a cluster-randomized controlled trial. Care managers offered a complex stepped-care intervention of monitoring, psychoeducation, and behavioral activation techniques. Twenty-six percent of the intervention group responds with a remission of depressive symptoms compared with 11% who received treatment as usual (TAU). The low-threshold intervention was more successful than TAU. Nevertheless, three-quarters did not respond with a remission. The aim of this study is to identify and describe the different types of utilization and of treatment response to understand what constitutes an effective intervention. Methods: Of 64 patients from the intervention group, we carried out problem-centered interviews with 26 patients from the intervention group. We analyzed the interviews using a qualitative type-building content analysis. For type construction, we performed a contrasting case comparison, regarding inductive and deductive categories of the intervention utilization and the symptom development. Results: The 26 participants' ages ranged from 62 to 87 years (mean = 72 years). Three participants were male. We identified five types of utilization, which differ primarily in the realization of pleasant activations, depending on own activity at the beginning and during the therapy: "activatable relief seekers," "active relief seekers," "active relaxation seekers," "passive problem-solving seekers," and "passive relief seekers." In the second typology, we analyzed four deductively determined types of treatment response responders, slight improvers, constant moderates, and non-responders. Patient-specific characteristics are a recent history of depression, an affinity for activities, supportive contacts, and limited comorbidity. In contrast, non-responders report contrary characteristics. Conclusion: Our two typologies emphasize that an effective intervention requires a match between intervention components and patient characteristics. We saw no intersections between utilization and treatment response. GermanIMPACT is an effective low-threshold intervention for moderately burdened patients, who are still capable of self-activation. An expansion of the intervention, especially for depression with a long history and comorbidities impairing mobility, could increase the effectiveness and improve the care situation of older people suffering from depression.