Dental Recall Metrics: Measuring Recall Without Vanity Numbers
Every recall program produces numbers. The question is whether the numbers you look at can tell you the program is failing. A surprising share of recall reporting cannot: it counts effort, effort always looks busy, and busy looks like success right up until the hygiene schedule says otherwise.
This article separates activity metrics from outcome metrics, defines the handful of numbers that actually describe a recall program’s health, and closes with the measurement discipline that keeps those numbers honest. For the program these metrics describe, see the guide to hygiene recall.
Messages sent is activity, not outcome
The most reported recall number is messages sent, and it is the least informative. Messages sent measures what the system did, not what patients did. A program can send more messages than ever while bookings fall, and a sent-message chart will show the whole decline as growth.
That is what makes it a vanity metric: it moves in a satisfying direction almost no matter what is happening underneath. Sent counts have a legitimate use as a health check, confirming the system is actually running, on schedule, without silent failures. Treat them like a pilot treats the fuel gauge: worth a glance, never the destination. The same caution applies to their cousins, patients enrolled and touches delivered. All of them describe the size of the attempt, none of them describe the result.
The numbers that matter, defined in words
Four outcomes describe a recall program, and each one is a checkpoint the patient either passes or does not. Define each in plain words before anyone builds a dashboard, because a metric the team cannot define is a metric the team cannot trust.
Replies. A reply is a patient responding to outreach in a way that opens a conversation: answering the text, calling the office and mentioning the message, clicking through to book. Replies measure whether the outreach lands with the right patients, at the right times, in words worth answering. If replies are weak, nothing downstream can be strong, and the fixes live in list quality, message timing, and what the messages say.
Bookings. A booking is an overdue patient with a new hygiene appointment on the schedule following outreach. Bookings measure the handoff: whether replies get answered while intent is warm and whether the conversation ends in a concrete appointment. Strong replies with weak bookings is not an outreach problem; that pattern points at the front desk workflow.
Kept visits. A kept visit is a booked patient who showed up and sat in the chair. This is the first number on the list that represents real production and a real resumed relationship. Reactivated patients booking far out or after a long absence need the same confirmation discipline as anyone else, and bookings that quietly die before the visit are covered in reducing hygiene no-shows.
Reactivations. A reactivation is a lapsed patient who is genuinely back: visit kept and the next one scheduled, returned to a normal recall rhythm rather than making a single guilt-driven appearance. Reactivations are the true output of the program, and their long-run value is the subject of reactivation versus new patients.
Read together, the four numbers form a funnel, and the funnel is a diagnostic instrument. Wherever the biggest drop sits between stages, that stage is where the program is leaking, and that is where the next improvement effort belongs. A single blended “success rate” hides exactly the information the funnel exists to reveal.
Attribution honesty: was the touch before the booking?
The subtlest way a recall dashboard lies is by claiming bookings it did not cause. The test is sequence in time: a booking counts as an outreach result only if a touch reached the patient before the booking happened.
Get this wrong and the errors compound. A patient books on their own on Monday; the scheduled recall text goes out Tuesday; a careless report credits the program with a win it had nothing to do with. Multiply that by every patient who would have returned anyway and the program looks better than it is, which means underperforming messages never get fixed, because the numbers insist they are working.
Two rules keep attribution honest. First, the touch must precede the booking, with no exceptions and no “close enough.” Second, the touch must be recent enough to plausibly matter: a patient who booked long after a single ignored message was probably moved by something else, and claiming them stretches the story. Where exactly to draw that recency line is a judgment call; that the line must precede the booking is not.
It is worth saying that honest attribution will make your program look worse than dishonest attribution would. That is the point. The purpose of measurement is not to flatter the program; the purpose is to find the weak stage while there is still time to fix it.
A cadence that keeps the numbers meaningful
Metrics only change behavior if someone looks at them on a rhythm and asks the same questions each time. A short recurring review, weekly or monthly to match the pace of your list, needs only the funnel and two questions: where is the biggest leak, and what changed since last time?
Resist the urge to redefine metrics mid-stream. A definition that shifts between reviews destroys the comparison that gives the review meaning. Write the definitions down once, in the plain words above, and change them rarely and deliberately.
Where CaseLift fits
CaseLift tracks the full funnel automatically, from outreach through replies, bookings, kept visits, and reactivations, by syncing with your PMS instead of relying on manual tallies. CaseLift applies before-the-booking attribution by default, so the wins on the dashboard are wins the outreach actually earned.