Digital technologies are being developed to address a broad range of health challenges. Mental health challenges may seem particularly suited for a digital solution. As mental disorders are experienced as behavioral or cognitive problems, digital phenotyping based on smartphone sensor data, keyboard performance, and voice/speech is especially promising for detecting and diagnosing anxiety, mood, or psychotic disorders. In contrast to traditional assessments which are subjective, episodic, and clinic-based; digital phenotyping can deliver objective, continuous, and ecological assessments of behavior. Because the assessment is passive and the technology is ubiquitous, there is understandable enthusiasm about this new approach to measuring behavior and cognition. But will better measurement result in better outcomes? We do not yet have the evidence to answer this question, although combining this afferent signal with the efferent use of online tools (from peer support to improved clinical management) may close the loop. This lecture will explore the promise and problems of digital phenotyping and online treatments, arguing that better measurement will be fundamental to better management of mental disorders and describing the studies necessary to prove the clinical value and ensure the public trust for this new technology.