Which channels produced patients who stayed?

A weight-management program earns over months. A patient who signs up and stops after four weeks is worth a fraction of one who stays a year, and channels differ enormously on that measure.

Ranking by program revenue over a longer range rather than by signups shows which channel produces patients who continue. It is common for the cheapest signup source to have the worst retention by a wide margin.

GLP-1 demand made inquiries abundant and cheap, which made cost per lead look excellent everywhere and stopped distinguishing between channels at all.

Revenue per acquired patient still distinguishes them sharply. That is the number that survives a demand spike.

In a medication-led program a substantial portion of what the patient pays leaves again to cover supply. Two campaigns can therefore produce identical revenue and quite different contributions, because one recruited patients onto a program tier with a high drug cost and the other filled a coaching and monitoring tier where almost everything stays in the clinic. Revenue alone ranks them as equals. The clinic's own bank balance does not, and the gap widens every month the wrong campaign is funded.

This is not an argument against ranking on revenue, which remains the right first instrument, but it is an argument for reading revenue per sale beside it. The channel ranking shows each source's sale count and revenue, so a book built on large numbers of low-value enrollments is visible as such rather than hiding inside a total. A program-tier column would make it explicit, but the program somebody is on is health information, and it stays out of a file sent to a vendor that signs no business associate agreements. Read on its own, revenue will eventually persuade a clinic to double down on the work it makes least money from, which is a hard mistake to reverse once the diary is full of it.

What the breakdown cannot show is which campaign feeds which tier, because it does not rank channels within a tier. If tiers turn out to be unevenly reached by marketing — the premium program mostly attributed, the entry tier mostly unexplained, or the reverse — that is already a decision about which tier the clinic wants more of, made on figures rather than on whichever advert produced the most telephone calls last week.

The commercial model here is a subscription wearing a clinic's clothes. A patient enrolling at $299 a month is worth $299 or $3,600 depending entirely on how long they stay, and the difference is settled somewhere around the third or fourth month, when the initial results have arrived and the novelty has not. Sources differ substantially in how many of their patients reach that point, and the differences do not correlate with what the enrollments cost to acquire. Enrollment counts treat the two patients as equal on the day they sign, which is the one day on which that is true.

Nothing available on the day of enrollment predicts it, which is why cost per enrollment has become the default metric and why it keeps producing disappointing quarters. A campaign that fills the diary in January and empties it by May looks excellent for four months and is then blamed on retention, the coaching team or the season. The pattern is only legible when each month's charge is credited back to the inquiry that produced the patient in the first place.

That is a matter of exporting twelve months of charges with the patient's contact detail on every row and running the report across the whole period. Recurring charges join to the same person and therefore to the same original source, so the ranking becomes revenue per enrollment rather than enrollments per dollar. It is common for the order to reverse outright, and for the source everyone had written off to be the one producing patients who are still paying in the autumn.

A local clinic with a scale, a phlebotomist and a physician in the building is competing for search terms against remote programs that operate at national scale and price accordingly. Those competitors can afford a cost per inquiry that would be ruinous locally, because their delivery costs almost nothing per additional patient. A clinic benchmarking its own cost per lead against what it assumes the market pays is comparing itself to a business with a different cost structure and drawing conclusions that do not apply to it.

The local advantage is real but it shows up later in the relationship rather than at the click: patients who come in, get measured, are seen by someone who remembers them, and stay longer as a result. That is a retention advantage and retention is exactly what cost per lead cannot see. So the local clinic looks expensive on the metric it can measure and is stronger on the metric that decides the year, and it usually cuts spend on the strength of the first.

Measuring on collected revenue per inquiry rather than cost per inquiry is what corrects the comparison, and it can be done for a period that has already closed because the billing records were kept anyway. A clinic that runs the last twelve months this way gets a defensible figure for what a patient from each source is genuinely worth, which is the only number that makes bidding against a national operator a decision rather than a nerve test. It also gives the clinic something to say internally when the cost per lead rises again, which it will, and the answer stops being an argument about whether the market has got worse.