What millions of appointments taught us about patient no-shows
A no-show looks like bad luck. In aggregate, across millions of appointments, it looks like something far more useful: a pattern. Once you can see the pattern, the empty slot stops being an act of fate and becomes something you can plan around.
#The lead-time curve is real
The single strongest signal is how far ahead the appointment was booked. A slot booked three weeks out is missed far more often than one booked for tomorrow — intentions decay with time. This alone reshapes scheduling: the further out a booking sits, the more a confirmation step earns its keep.
#Time of day matters more than you'd think
Early-morning and late-afternoon slots are missed at noticeably higher rates than mid-morning ones. Mornings collide with commutes and school runs; late slots lose to the day running long. Knowing this, you can place your most reschedule-tolerant appointment types at the riskier hours and protect the scarce, hard-to-fill specialist slots for the windows people actually keep.
#Reminders work — but timing and channel decide how well
Reminders reliably cut no-shows, but a single reminder a week early does little. What moves the needle:
- A confirmation at booking, so the appointment feels real.
- A reminder 24–48 hours before, with a one-tap way to confirm, reschedule or cancel.
- The channel the patient actually reads — which is rarely the one that's cheapest to send.
The cancel option matters as much as the reminder. A patient who cancels at 24 hours hands you a slot you can refill. A patient who simply doesn't show hands you nothing.
A cancellation is a gift. A no-show is a loss. Good reminders convert the second into the first.
#History is the best predictor
Past behaviour predicts future behaviour more reliably than any demographic factor — and leaning on history keeps you away from the ethical minefield of profiling by age, gender or location. A patient who has missed two of their last three appointments is genuinely high-risk; treat the risk, not the person.
#What to do with a risk score
Once a model assigns each upcoming slot a no-show probability, you have options that don't punish anyone:
- Gentle overbooking of high-risk slots, calibrated so a rare double-arrival is manageable.
- An extra reminder for the highest-risk bookings.
- Waitlist offers that automatically fill a slot the moment it's released.
The goal isn't to shame patients into attending. It's to keep capacity full while making it as easy as possible for people to keep — or cleanly release — their appointment.
#The quiet payoff
Hospitals that take no-shows seriously don't just recover lost revenue. They shorten waiting lists, because reclaimed slots go to someone who needs them, and they improve the patient experience, because the reminders and easy rescheduling are genuinely helpful. The data was always there in the appointment book. It just took looking at enough of it to see the shape.