How to Tell If It Worked

Last week I argued that if a practice is going to try AI anywhere, it should start with the task where it can actually see the result. Then I did not say how to see it. That was a hole in my own argument, and this post fills it.

Here is the constraint I am writing under. AiroDental does not publish outcome figures, because every number we could publish would describe some other practice's phone, schedule, and definition of a new patient, and an earlier post on this blog was a whole argument about why those definitions do not travel between practices. So what follows is not a benchmark. It is a way for a practice to measure its own front office, before and after any change, using nothing it does not already own.

The ten numbers, and the one that is missing

The ADA's Guidelines for Practice Success has a page on key performance indicators. It lists ten commonly tracked: production, collections, profit, overhead, new patients, fee-for-service versus insurance production, case acceptance, doctor versus hygiene production, the percentage of hygiene patients scheduled, and the cancellation and no-show rate. The page suggests starting with a few and reassessing at least quarterly, and notes that many "can be found in your practice management software."

Now notice what is not on it. Not one of the ten measures the phone. Every number on the ADA's list describes what happened after a person got through to the practice: whether they were seen, accepted, paid, came back. Nothing on it describes whether they got through at all. The list starts at the front desk and looks inward, and the front office's first job happens before the list begins.

That is not a criticism of the ADA. The phone has always been the one instrument in the building that kept its own records without anyone reading them.

The log nobody opens

Every business phone system keeps a call log. For each call it records the time, the number, how long it rang, whether someone answered, how long the conversation ran, and, on most modern systems, whether the call went to voicemail. If your phones are on an internet-based service, the log is a report you can open in a browser. If they are older, it is still there, and the vendor can pull it.

Most practices have never looked. The log was built for billing, not management, and it does not come with a summary. But it holds three things a practice cannot get anywhere else: the calls that rang out, meaning nobody picked up and the caller left no message; the calls that came in while the office was closed; and the length of the calls that were answered, which tells you what the scheduling coordinator was doing when the next line lit up.

Pull one day. Not a special day, just a day. Sort by outcome. Count the calls that rang out during business hours and the calls that came in after them. Those two numbers are the front office's denominator, and no KPI page has ever asked for them.

What the schedule already knows

The other half lives in the practice management system, and the ADA's list points at it: new patients, and patients by referral source.

Every system reports new patients, and each defines them a little differently. Open Dental's New Patients report, for instance, counts a patient as new on the date of their first completed procedure, not their first visit, and its referral column shows only the first referral source entered. That is a fine definition, and also a reminder that the "how did you hear about us" field is only as good as the person who filled it in. If the desk types "phone" for everyone who called, the report will say everyone came from the phone.

So before you measure anything, decide what the desk writes in that field, and write that down where the newest hire will find it.

Put the two halves together

This is the whole method. Take the day you pulled from the phone log and list every number that called for the first time; your phone system can tell you which numbers it had never seen before. Then, a week or two later, look each of those numbers up in the practice management system. How many are on the schedule?

That fraction is the measurement: first-time callers who became scheduled patients, over first-time callers. You built it in an afternoon, from two systems you already pay for, and it describes your practice and nobody else's. Run it before any change to the front office and again afterward, the same way, and you will know whether the change worked in the only sense that matters.

Do the same for the after-hours calls. How many got a return call the next business morning, and how many of those callers ended up on the schedule? That is a smaller list and a harder look, because in most practices nobody owns the after-hours list, which means the number is not "low," it is "nobody knows."

What changes when it works

If a front office change is working, three things move, in this order. The ring-out count in the phone log goes down first, because that is what is being fixed directly. Then the after-hours list gets a next-day return every time, because someone or something now owns it. Then, later, the new-patient count on the ADA's list moves, because the people who got through are on the schedule where the KPI page can finally see them.

If the first number does not move, nothing downstream will, and you should stop and ask why. If the first moves and the last does not, the problem has moved inside the practice, to the handoff between the phone and the schedule, and that is a different fix.

What it will not tell you

Two lists will not tell you whether the caller felt heard, or whether the treatment coordinator's tone on a financial call sent someone to another office. For that there is no shortcut. Listen to the calls. Most phone systems can record them, with the disclosures your state requires, and a practice that listens to a handful of its own calls learns more about its front office than any report will show.

If the practice uses Laine, the listening is easier than people expect, because the number you receive at signup is an ordinary working number that is not published anywhere. Until you connect it to your phone system, your office is the only place that has it, so you can call it yourself, as a patient would, and hear exactly what a patient hears before a real call reaches it. Laine answers the calls that would have rung out and books the caller into the schedule through the practice management system, and the phone log is where you will see whether that is happening, in the ring-out count you were already measuring. It works today with Dentrix, Dentrix Ascend, Open Dental, and CareStack.

We will not tell you what your fraction should be. We will tell you how to find it, and then you can hold us to it. Book a walkthrough once you have run the two lists. There is no long-term contract and no onboarding fee, so if the second run does not move, you are free to say so and leave.

Deren Flesher, DDS. Founder, AiroDental.