Why we said no to scaling prescription mills

Three years in, we sat through a board paper that recommended a major shift in our prescribing model across one of our clinical programs. The argument was simple: lift the prescribing rate by a meaningful margin and the business added several million dollars of recurring revenue inside twelve months. The patient already wants the prescription. They’ve paid. They’ve completed the questionnaire. Why are we saying no to a meaningful share of them?

We said no to the change. Here’s the longer answer.

What our clinicians were declining, and why

When we audited the cases where our clinicians declined to prescribe, the reasons were not arbitrary. Across our brands, the most common were:

• Untreated psychiatric comorbidity that needed addressing first.
• Pregnancy or pregnancy planning in the relevant window.
• Active disordered eating patterns or another condition where another intervention is more appropriate.
• Patient goals not aligned with what the treatment does (someone wanting a cosmetic outcome the clinical pathway isn’t designed for).
• Interactions or contraindications the patient hadn’t disclosed elsewhere.

Each of those is a clinical decision a thoughtful prescriber would make in person. The job of the platform is to make sure they CAN make that decision, not to override it.

The funnel-versus-clinical tension

The way most digital health products are measured, declining to prescribe is a “drop-off.” A funnel metric goes red. Someone in growth wonders why we’re paying for the marketing if we’re not converting.

We built our clinical programs on a different premise: the treatment is a tool, not the product. The product is the program. A patient we decline today might come back in three months ready, or might never need to. Both are fine outcomes.

How we made it stick operationally

Stating values is easy. Making them load-bearing requires structure:

1. Decline-to-prescribe is a logged, audited clinical decision with its own coded reason. We can pull a report on it any week.
2. Clinicians are paid by hour, not by script. This removes the most obvious incentive distortion.
3. The Clinical Council reviews prescribing patterns monthly. Any clinician whose prescribing rate is two standard deviations outside the cohort gets a conversation, in either direction.
4. We don’t run growth experiments on prescribing thresholds. Pricing, copy, channel mix, yes. Whether to prescribe, no.

The thing nobody benchmarks against

There is no industry benchmark for “appropriate decline rate.” We know what ours is across each program, and we know what we’d be uncomfortable with. That’s enough. If you’re building a clinical service and you can’t tell anyone what your decline rate is, that’s the first metric to start instrumenting.

The clinician-payment structure behind this is its own piece: pay clinicians by the hour, not the script. And for the wider framework, see our clinical governance page.