No-show rate
A rate that counts three different events, costing a session, a phone call and nothing respectively.
Also called DNA rate, Did not attend rate, Missed appointment rate.
What it is
The share of booked appointments where the patient or customer did not attend and the slot was not recovered. The words "and was not recovered" are the whole metric, and they are what most systems do not record.
unrecovered missed appointments ÷ appointments offered
- unrecovered
- Nobody used the slot, so the capacity was genuinely lost
- missed
- The patient did not attend, excluding changes you initiated
- offered
- Slots you made available, from the rota rather than from bookings
The part the formula leaves out
The status field is doing three jobs. "Did not attend", "cancelled late and we filled the slot" and "we moved it ourselves" are usually written into the same column, and they cost a session, a phone call and nothing. Separated, the worked example's 22.4% becomes 11.4%, and the smaller number is the one you can act on.
The denominator is the second problem. Dividing by appointments booked cannot exceed 100% and says nothing about capacity; dividing by slots offered tells you what share of the capacity you paid for went unused. Those are different metrics answering different questions, and the second is the one that shows up in the accounts.
The predictor is usually not in the file people look at. Booking lead time is the strongest variable in most appointment data and the one most exports bury: in the example, appointments booked more than 60 days out are missed at 3.9 times the rate of those booked within a week, monotonically across every band, holding inside every clinician and every appointment type.
How it is usually computed wrongly
01
Counting late cancellations as no-shows
A slot cancelled with notice and refilled cost nothing. Including it roughly doubles the rate and makes the metric unusable for capacity planning.
Count only what was genuinely unrecovered, and report the recovered ones separately since they measure how well your waiting list works.
02
Dividing by booked rather than offered
It hides empty capacity entirely. A session half booked and fully attended looks perfect.
Take the denominator from the rota, excluding blocked administrative time, and keep walk-ins out of it since they occupy a slot that was never offered.
03
Reporting one rate for the whole practice
It averages a Saturday clinic at 96% attendance with two midweek afternoons in the sixties, and recommends nothing.
Split by session, day, appointment type and lead time before drawing any conclusion. The variance is the finding.
04
Judging reminders on whether they were sent
Reminder effectiveness depends on timing relative to when the booking was made, not on volume. Sending more of them at the same fixed offset changes nothing.
Measure attendance by reminder timing, and specifically for long-lead bookings where the reminder is doing the most work.
What your file needs
- One row per appointment with an outcome status
- The date the appointment was booked, not only the date it was for
- A rota or session template, for the offered-slot denominator
Anything missing is reported as unavailable rather than substituted with something weaker computed on worse evidence.
Compute it on your own file
No account needed to start. You only pay when you like what you see.