Healthcare marketing attribution and reporting

Every figure we report traces to a named system: the practice management system, the phone system and call tracking, the booking platform, the ad platforms, the general ledger. Where a number is derived rather than measured, we publish the arithmetic that produced it and name the inputs, so it can be argued with rather than believed.

Then the limits, because they are real and every competitor’s dashboard pretends otherwise. A patient who saw a sign, asked a colleague and typed your name into a browser three weeks later is not attributable to a channel by any honest method, and we will not assign her to one. Cross-device paths break. Platform-reported conversions are modelled by the platform and are not the same population as your booked patients. Offline conversations are invisible unless somebody records them. Any view of the journey assembled after the fact is a reconstruction.

So we report two things and keep them apart: what is traced end to end, and what is estimated with the method stated. The first is usually a smaller number than an agency dashboard shows, and it is the only one worth setting a budget on.

A new patient does not appear out of one advertisement. She circles — a search, a map, reviews, a friend, a second search — over weeks, and only the last of those touches leaves an obvious fingerprint. In a naive setup that last touch takes the entire credit, and the budget is then rebuilt around it.

The consequence is specific and expensive: branded paid search, a click you paid for to reach somebody who already knew your name, reports as your best channel, and the six touches that actually created the patient are quietly defunded.

The full worked example is published — one patient, nine touches, the same revenue assigned two ways, and the two opposite budgets that result.

Point an ad algorithm at "conversions" without telling it which conversions are new revenue, and it will do exactly what you asked: find the cheapest, most certain conversions available. Nothing is cheaper or more certain than a patient who has already decided to call you and is searching your name to find the number.

The loop is airtight and pointed the wrong way. Cost per conversion falls, the dashboard improves, spend rises to match — and the bank account does not move, because the budget is buying visits that were going to happen anyway.

This is the second claim on this site a reader can verify without trusting a single ADMEN figure: open your own account and look at what share of conversions came from searches containing your own name. The full mechanism of the branded-search trap, the correction, and what a holdout test costs are published.

Monthly, because a week is too small a sample to make a decision on — and because nobody reads the weekly one.