Every clinic counts new patients. Almost none can count the ones who booked.

Every clinic group I've worked with can tell you how many new patients they had last month. Almost none of them can tell you how many of the people who booked an appointment actually became patients.

Those sound like the same question. They are not, and the gap between them is where most clinic marketing budgets quietly go to die.

Two systems, one missing key

The pattern is nearly universal in dental and aesthetic medicine. There's a practice management or agenda system that owns appointments — who booked, for what, with whom, at what time, whether they turned up. And there's a transactional record that owns money, keyed on a patient number.

Reporting almost always gets built on the second one, because money is unambiguous and patient numbers are clean. Count distinct patient numbers appearing for the first time in a month and you have "new patients". It's a real number. It's just a rear-view mirror.

What it can't tell you is anything about the journey in front of it. How many booked consultations turned into a first treatment. Which channel produced people who came back for the second and third appointment rather than the ones who booked once and vanished. Whether the campaign that generated forty bookings generated four patients or thirty.

All of those questions need one thing: a shared key that joins the appointment record to the transaction record. If the agenda doesn't stamp the patient number at the moment of booking, no dashboard will ever recover it. You cannot join two tables on a column that isn't there.

Why marketing then optimises the wrong thing

Give a marketing team bookings as the only measurable outcome and they will get very good at producing bookings. That is not cynicism, it's a completely rational response to the feedback they're given.

So the ad platform gets taught that a booking is the goal, and it goes and finds more of whatever produces bookings — including the people most likely to book on impulse and least likely to show up. The bidding algorithm is doing exactly what it was told. Nobody told it the difference between a booking and a patient, because nobody could measure it.

The technical fix is unglamorous and well documented. Google's own offline conversion import exists precisely for this: you store the click identifier alongside the lead, and later — when you know whether that person turned into anything — you send the real outcome back. The mechanism has existed for years. What blocks it is almost never the ad platform. It's that the clinic's two systems don't share an identity.

David McRaney has a line in You Are Not So Smart that explains what fills the vacuum: you don't think in statistics, you think in examples, in stories. When the join is missing, a management meeting runs on the three patients somebody remembers.

Identity is harder in healthcare than anywhere else

It's worth being fair about the difficulty, because this isn't sloppiness. Patient identity is a genuinely hard problem, and one that the health sector has been fighting for decades.

The Pew Charitable Trusts' research on patient matching in electronic health records found that as many as one in five patients may not be correctly matched to their existing record within a single organisation — and that the failure rate can approach half when records move between organisations. Matching mostly relies on demographics: name, date of birth, address. People marry, move, shorten their first name, get typed in wrong at reception.

The standards world has an answer for the structure, if not for the human mess. The FHIR Patient resource exists so that a patient identity can be referenced consistently across systems instead of re-derived from a name each time. Most clinic software can speak some version of it. Very few clinic groups have decided which system is allowed to mint the identity in the first place.

That decision — which system owns identity — is a management call, not an IT call. It's the one that has to happen before any of the reporting work is worth starting.

The sequence I'd run

Pick the system of record for patient identity and say it out loud. Then make sure the identifier is written at the earliest possible touchpoint, which is the booking, not the invoice. Then join the two datasets and look at the difference between bookings and patients before you build anything pretty on top.

Only after that does it make sense to feed real outcomes back into the channels. This is the same problem I've written about in commerce, where merging four sales channels into one number was less about the merge and more about proving the number was right afterwards — and the same reason brand search flatters itself in every dashboard that measures the last step rather than the whole journey.

Healthcare just raises the stakes on both ends. The margin per patient is high enough that getting the join wrong wastes real money, and the subject matter is personal enough that the marketing has to be careful in ways most categories never have to think about.

Count patients, not bookings. If your systems won't let you, that's the project — not the campaign.

Sources & further reading

External
Enhanced patient matching is critical to achieving full promise of digital health records — The Pew Charitable Trusts.
Patient resource — HL7 FHIR specification.
About offline conversion imports — Google Ads Help.

Related posts
I merged four sales channels into one number. Then I proved it was right.
Your brand search isn't the hero your dashboard thinks it is
The high-stakes art of selling a smile
The boutique hotel-ification of healthcare