Tuesday, 10am. Your first patient hasn't arrived. At 10.12 you ring: voicemail. At 10.20 you write it off, and nobody else is in until eleven. Forty minutes of a room that costs the same empty as full.
It happens once or twice a week. It almost never gets written down.
The figure everyone quotes has no study behind it
Search for what a missed appointment costs a private clinic and you get the same range everywhere, usually rendered in pounds: 80 to 250 per appointment. Trace it and you land on practice-management software blogs. One states that a missed appointment "typically represents between $80 and $250 in lost revenue" and cites nothing — no survey, no sample, no method. It's a plausible guess at a consultation fee, dressed as a finding, published by a company selling the fix. The Spanish version of the same press cycle is blunter still: a supplier's release in September 2025 put clinic losses at €2,500 to €7,500 a month and claimed its own product cuts no-shows "by up to 50%", sourced to "studies published in Elsevier" with no title and no sample.
There is no published figure for UK private practice. There is good public data, and it needs dating properly. NHS England reported on 2 March 2026 that 16 million GP appointments went unattended in 2025, out of 376 million arranged — around one in 23. Hospital outpatients is a different animal: of 146.1 million outpatient appointments in 2024-25, 77.5% were attended, the rest split between did-not-attends and cancellations by patient or hospital. Between 2023-24 and 2024-25, DNAs rose 1.5% and patient cancellations rose 18.0% (NHS England, Hospital Outpatient Activity 2024-25).
Don't drop any of those into your own P&L. Where the appointment is free at the point of use, missing it costs the patient something different.
Your number comes from three columns on a spreadsheet
Separate the no-show from the late cancellation. Someone who texts at eight about a ten o'clock slot leaves you something to re-offer. Someone who simply doesn't turn up doesn't. Blending them inflates the figure and hides which fix you need.
Record whether the slot got filled. A no-show with a waiting list costs one phone call. A no-show at ten on a Tuesday in August costs the whole session.
The cost isn't the fee. It's the fee minus what you'd have spent seeing them: materials, lab work, the clinician's cut if they're paid per session.
Multiply the month's unfilled no-shows by the average margin on that slot. That's your number.
What the trials actually say about reminders
There is real evidence here, and it's more modest than the sales pitch.
The most recent systematic review (Al-Turbag, Mooney and Corry, Journal of Hospital Management and Health Policy, 30 March 2026; 10 trials, 8,236 participants) found reminders lifted attendance by 11% in relative terms (RR 1.11, 95% CI 1.05–1.19). Split by channel, the surprise: telephone reminders were significant (RR 1.11, CI 1.04–1.19) and SMS on its own was not (RR 1.14, CI 0.99–1.31, with 91% heterogeneity). The authors note that none of the included studies costed the saving.
There's also an arithmetic trap. That 11% was measured in clinics with poor attendance. If you're already at 90%, there is no 11% to win: the whole ceiling is ten points, and reminders won't take all of them. In absolute terms the larger trials move two or three percentage points. That's what to expect.
The 2013 Cochrane review (Gurol-Urganci et al., CD007458) adds the practical bit: SMS and phone calls performed the same (RR 0.99, CI 0.95–1.03) and SMS cost less per attendance gained. Which dismantles the vendors' favourite line. That "98% open rate on WhatsApp versus 20% for email" is attributed to Meta, we found no independent verification of it, and no trial exists comparing WhatsApp against SMS head-to-head for appointment attendance. A message being opened more isn't the same as a patient turning up more.
The thing that genuinely moved the needle was the wording
In 2015 Michael Hallsworth's team published two randomised trials in PLOS ONE, run at an English hospital with 10,111 and 9,848 patients. Same channel, same timing — SMS five days out — different text.
The standard reminder left an 11.1% did-not-attend rate. The one naming a specific figure got it to 8.4%: "Your hospital appointment is on [date] at [time] at [location]. If you don't attend, it costs our NHS about £160. Please call [number] to cancel." Saying the same thing vaguely — that missing costs the health service money — only got to 9.9%. Specificity was the active ingredient.
And here's the honest part: that message doesn't transfer to private practice. It worked because the NHS is free at the point of use and the patient was learning about a cost they hadn't considered. Yours has already paid. Reminding them the session is £70 tells them nothing. Naming the person on your waiting list who loses the slot instead sounds sensible, but nobody has tested it in a trial. If you try it, run two groups.
The same study produced one finding that does transfer: the arm that only made cancelling easier fell from 11.1% to 10.1% without saying anything else. A full point for putting the button within reach.
Let them cancel in one tap (and why that's also cheaper)
NHS England says it explicitly in its guidance on reducing DNAs: reminders work better where the patient can reply to cancel. Its March 2026 campaign runs on the same logic — notifications in the NHS App so people can cancel or rearrange.
There's a billing detail almost nobody explains. On WhatsApp, a reminder for an appointment booked three weeks ago is a paid utility template: it falls outside the customer service window. But the moment the patient replies, even with "ok", a 24-hour service window opens, and Meta's documentation is blunt: "All non-template messages are free", and utility templates delivered inside an open window are free too. A two-way reminder doesn't just work better, it makes everything after it cheaper. We went through the mechanics in WhatsApp Business API.
One legal line worth knowing. The ICO treats a neutrally worded appointment message as a service message, not direct marketing, so it needs no PECR consent. Add a discount code or a "book your hygienist while you're here" and it becomes marketing, which does — and Meta will reclassify the template as marketing, at a higher rate. Expensive on both counts.
What reminders don't fix
They don't make people come who don't want to come. In the Censuswide survey for NHS England of 2,000 adults in England (2 March 2026), 12% said they had forgotten an appointment and 11% that they had arrived too late. Forgetting is one cause among several. Against fear of the chair, the price, or the parking, a text can do nothing.
They don't fill the slot. Learning that someone isn't coming and getting someone else in are two different jobs. We covered that here.
They don't replace a cancellation policy, which has its own limits under the Consumer Rights Act.
They don't assess anything clinical. A reminder confirms or cancels. The moment a patient replies describing a symptom, that's a person's job.
Measure it for a month before buying anything. If it comes to two empty slots, you have your answer.
Zatio does the answering part: it handles inbound WhatsApp and web chat enquiries, confirms appointments and takes the cancellation when a patient writes in, whatever the hour. It doesn't triage, doesn't decide who gets the freed slot, and says it's an AI in the first message. We have two clients and neither is a clinic, so we're not going to show you clinical results we don't have.