The 4:55 on Friday test
A ten-minute exercise that tells you more about what your front desk is costing you than any consultant will. You need a phone and a stopwatch.
By Devin Picciolini — Founder and CTO of Slate. Built a healthtech platform used by 25,000 patients and sold it. Ten years shipping software in regulated industries.
Every practice owner I talk to has a confident answer to the question "how are your phones?" The answer is almost always "fine, busy, we manage." And almost every one of them is wrong, not because they're careless, but because nobody has ever measured it. They're describing the phones from inside the building, and the only opinion that matters is from outside it.
Here is a test that costs you ten minutes and no money at all. Do it before you spend a dollar with anyone, including us.
The test
Pick a Friday. At 4:55pm, call your own main number from a mobile number nobody there recognizes. Don't warn anyone. Start a stopwatch.
Then write down four things:
- How long until a human speaks. Not until it rings — until an actual person says an actual word.
- What the caller hears in between. Hold music? A menu? Six rings and voicemail? A recording that tells them the office hours they're currently calling during?
- Whether you could have booked an appointment. Not whether you got through. Whether a brand new patient, with money, could have got themselves onto the schedule right then.
- Whether anyone would ever have known you called. If it went to voicemail and you hung up, does that call show up anywhere at all?
Now do it again at 12:30 on a Tuesday. And once at 8:10 in the morning, before the doors open.
Why that fourth question is the important one
Most owners look at the first three and feel about how they expected: a bit of hold time, an occasional voicemail, nothing scandalous.
Number four is where the money is.
A call that rings out and doesn't leave a voicemail leaves no trace in your practice management software. It does not appear in any report you have ever looked at. It does not show up in production, in new-patient counts, or in your monthly numbers — because in your numbers, it never happened. The patient exists, they wanted to give you money, and from your side of the wall there is no evidence they ever tried.
This is the only category of business loss with no reporting line attached to it. Which is precisely why it goes unfixed for a decade in practices that are otherwise run beautifully.
Do the arithmetic with your own numbers
Don't take an industry statistic from a vendor's landing page. Use figures you can defend:
- A. What is a new patient worth to you over the first two years? Your average, from your own software. Most practices know this or can get it in ten minutes.
- B. How many calls does your phone system report as missed or abandoned in a month? Your VoIP provider has this. If you're on a system that can't tell you, that's its own finding.
- C. What fraction of your inbound calls are new patients rather than existing ones? If you don't know, ask the front desk for a guess and use the low end.
Then: A × B × C = the monthly number.
Even being deliberately pessimistic on every input, most practices that run this calculation for the first time arrive at a figure that is larger than a full-time salary. That's the reaction I'm looking for when I ask people to do this — not because it justifies buying anything, but because it moves the phone from "an annoyance" to "the largest unmanaged line item in the business."
What people do next, ranked from worst to best
Worst: nothing. Which is the most common outcome, because the number is uncomfortable and the fix sounds like a project.
Bad: hire another front desk person. This works, sort of, at roughly $45,000 a year plus benefits and management overhead, and it does nothing at 7pm, on a Sunday, or during the four minutes when both people are with patients. You've bought coverage for the calls you were already getting, not the ones you weren't.
Better: an answering service. Cheap, immediate, real humans. But they can't see your schedule, can't answer whether you take a specific insurance plan, and can't do anything except take a message — which means the call still isn't booked and now the patient has waited twice.
Best: something that picks up every time and can actually book. The argument for an AI receptionist is about availability rather than technology. The technology matters only because it happens to be the first thing that can answer at 4:55 on a Friday, at two in the morning, and during the lunch rush, at a cost that barely moves with call volume.
The trap on the other side
Having said all that: an AI that answers every call and books nothing is worse than voicemail. You've replaced a patient who knows they didn't get through with a patient who thinks they did.
So if you take one thing from this: the test above measures availability, and availability is only half the problem. The other half is whether the thing that picks up can actually finish the job — get into your schedule, know which providers do what, know which insurance you accept, and know when to stop talking and get a human.
That half is boring integration work, and it's the half most vendors gloss over in the demo. It's also the half that decides whether the thing is worth anything at 4:55 on a Friday.
Run the test. If the number bothers you, the scorecard will place it against the rest of your operation in about two minutes, or the audit will put a defensible figure on it in two weeks.