Changing behaviour at scale

Technology has changed how behaviour change is delivered. A programme that once needed a clinician or a classroom can now reach millions through a phone: a weight-loss course, a smoking-cessation app, an automated text, a wearable that counts steps and prompts its wearer to move. Health services commission them, and employers and insurers offer them. Delivery at scale is no longer the obstacle. Whether behaviour changes once they are delivered is.

The trials are more encouraging than the populations. Digital programmes often match face-to-face ones in the trial that tests them, then lose most of their users within weeks of being rolled out. Effects measured in the motivated few who stay rarely carry to the many who sign up. Those with most to gain (older, poorer, less healthy people) are often the least likely to use them, so a programme can widen the gaps it was meant to close. Personalisation, from feedback on a wearable to an AI coach, is the newest promise; the evidence on it is still mostly short trials in volunteers.

The evidence does not yet show durability or transfer. It is not established whether effects measured over months hold over years, or whether a programme that worked in one health system works in another. Replications are rare, the trials short, and null results under-published.

Trackers