Why human connection makes digital support stick

Attrition is structural, not personal

Digital health has a well-documented drop-off problem. It was named the law of attrition two decades ago, and the shape has barely changed since: even carefully designed programs see substantial non-usage and dropout, and they see it early.

It is tempting to read that as a motivation failure. The more useful reading is that an unattended program competes against fatigue, stress, illness, travel, and the ordinary friction of a week, and it competes alone. Nothing is waiting on the other side of a missed day.

The 51/49 principle

Our shorthand for what the evidence keeps showing is that lasting change needs a slight edge of human connection. Not a majority-human program, and not a purely digital one. A little over half of the weight has to rest on a person.

The digital layer is what makes the support affordable, consistent, and available at 11pm. The human layer is what makes it matter that you showed up.

Supportive accountability

The mechanism has a name in the literature: supportive accountability. Adherence to an eHealth program improves when a person is involved, that person is seen as trustworthy and benevolent, and expectations are set collaboratively rather than imposed.

The detail that matters most for design is that it is the kind of contact, not the amount. More notifications do not substitute. A coordinator who knows your name and expects to see you again does.

Guided beats unguided, consistently

Across digital intervention research, supported versions outperform unsupported versions of the same program. The gap tends to widen for people with more severe presentations, which is precisely the group a weight-management program cannot afford to lose.

Meta-reviews of digital mental health reach the same conclusion, while noting that support is defined inconsistently across studies. The direction is stable even where the magnitude is not.

The mechanism is not domain-specific

This is not a mental health finding that happens to be interesting to us. The same pattern appears in habit formation, sleep, nutrition, movement, medication adherence, and recovery. Whenever effort rises and isolation returns, engagement falls.

Therapeutic alliance also appears to survive the move to digital settings, though its relationship to outcomes may work differently than it does face to face.

What this changes in practice

A handful of design commitments follow from it, and all of them are visible in how trei is built.

• Design for re-entry, not for perfection. A missed week should have an obvious way back in.
• Make support feel benevolent rather than clinical. Accountability without warmth reads as surveillance.
• Use digital touchpoints between human moments, not instead of them.
• Deliver help when it is timely, rather than making constant demands on attention.