Which channels produced admissions?

Cost per call in this category can run into the hundreds. At that price the difference between a channel producing calls and a channel producing admissions is the difference between a viable program and an unaffordable one.

Ranking on completed admissions rather than call volume regularly reorders the channels entirely — and the cheapest calls are frequently from the audience least likely to admit.

A large share of admissions come from clinicians, previous patients, and families who found the program some other way. That belongs in Direct / Unknown rather than credited to whichever campaign was running.

Its size matters. If half of admissions arrive that way, the marketing question is different from what a paid-channel report alone would suggest.

Two clients admit on the same Tuesday from two different campaigns. One completes detox and leaves after six days. One steps down through residential and a partial program and is in treatment for seven weeks. On any report counting admissions those are one each. In the accounts they differ by a factor of six, and in the census they differ by whether the building was full in March. A provider funding its marketing on admissions is funding it on the smallest and least informative fact about each client.

Length of stay is not random across sources either. Callers who arrive through a campaign promising immediate placement behave differently from those who found the program through a clinician, a family member who researched for two weeks, or an alumnus. The variation is large, it is stable enough to plan around, and it does not appear anywhere in an advertising account, which closes its books on the day the call connected and never revisits the record. Providers who have measured it describe the spread between their best and worst sources as being wider than the spread between their levels of care, which is not a small claim.

Exporting billed days or episode totals with a client contact detail on each row puts the revenue beside the source. Every row for one client matches that client, so a seven-week stay billed in several rows credits all of its revenue to the inquiry that produced it; each row still counts as a sale in the report, so episode totals keep the sale count meaningful. That moves the argument from lead quality, which nobody can settle, to collected revenue, which nobody can dispute. Keep clinical detail, including level of care, out of the file.

This segment has a matching problem most do not. The call comes from a mother, a spouse or an employer, and the client whose name eventually appears on the billing record is somebody else entirely. A join on phone numbers alone will miss that connection, and a system willing to guess at it would be inventing a link between two people on the strength of a shared surname or an overlapping date, which is not evidence of anything. So the single most valuable field in the whole exercise is one that is usually typed into a free-text note and never exported at all.

The design decision here is to refuse the guess. Only exact email and phone matches count automatically toward the revenue figures. The one weaker link the engine produces — the same national number under a different country code — is held for review, and a caller and a client with different contact details are simply not linked at all. That produces a lower headline match rate than a tool willing to be optimistic, and it produces a number an operator can put in front of a board without qualifying it, which is the trade this product makes everywhere and makes most visibly here.

Intake is where this gets solved, not in the software. Where the admissions team records the client's own phone or email on the inquiry row, alongside or instead of the caller's, the lead file carries a detail the billing record also has, and the join succeeds cleanly. Many providers already collect this; it sits in a call note rather than in a column, and moving it into a column is a smaller change than it sounds and the single largest improvement available to the match rate in this segment. The admissions team is already asking the question; the change is where the answer is stored.

Between the inquiry and the admission there is a verification step that decides whether the client can be accepted at all, and it filters unevenly by source. Campaigns written around cost and coverage attract callers whose coverage is the deciding question, and a proportion of them will not verify at a level that supports the program. The call happened, the admissions team spent forty minutes on it, and nothing followed. Counted as a call, it looks identical to the one that admitted for seven weeks.

At the prices this category pays for a click, that filter is the difference between a viable acquisition cost and an indefensible one. A source producing calls that verify at half the rate of another is effectively twice as expensive per admission, and no dashboard inside an advertising platform contains the data to notice. Providers generally feel it as a complaint about lead quality long before anybody produces a figure for it. What they have instead is an impression formed by whoever happened to take the difficult calls that week, which is not a basis for moving six figures of annual spend.

The figure comes from uploading both files. Calls and inquiries on one side, admitted and billed clients on the other, and the ratio per source falls out without any modeling. What a provider can do with it is not to switch the weaker channel off but to change what it says and where it qualifies, because the volume was real and the filtering was simply happening after the expensive part instead of before it. It is also worth running over a period that has already closed, since the inquiry logs and the billing records both exist and neither of them needed anything installed at the time.