How to cut no-shows in a private clinic, with the evidence attached.

A missed appointment costs a private clinic twice. The obvious loss is the fee. The larger loss is the clinician’s hour, which was paid for whether or not anyone sat in the chair, and which cannot be sold again once it has passed. Most clinics count the first loss because it shows up as a gap in the ledger, and ignore the second because nothing shows up at all.

The scale of the problem in the NHS is a reasonable proxy for what a private practice faces, and it is measured properly. In 2024-25 there were 146.1 million outpatient appointments in England, of which 113.2 million were attended and 8.1 million were recorded as did not attend, a DNA rate of 5.6 per cent. That is down from 5.9 per cent the year before and 6.2 per cent in 2019-20, but the absolute number still rose by just over 123,800. When NHS England launched its AI pilot at Mid and South Essex in 2023, the trust’s baseline DNA rate was 8 per cent, and NHS England put the national cost of missed hospital appointments at £1.2 billion a year.

Private clinics rarely publish their rates, and the honest answer is that yours depends on the specialty, the lead time and whether the patient has already paid. The figure to track is not the no-show rate alone but revenue per available clinical hour, because that captures the missed fee, the wasted hour and the slot you could have refilled if you had known in time.

The evidence on reminders is better than the evidence on almost anything else a small clinic can buy, because it has been randomised and pooled several times. The largest UK-authored synthesis is a BMJ Open meta-analysis from King’s College London, covering 21 studies and 16,076 patients. Patients who received electronic notifications were 23 per cent more likely to attend (67 per cent against 54 per cent) and 25 per cent less likely to no-show (15 per cent against 21 per cent). Multiple notifications were significantly more effective than a single one, and voice notifications appeared more effective than text.

The Cochrane review of mobile phone reminders, eight trials and 6,615 participants, reached a similar conclusion with a useful twist. Text reminders beat no reminder (risk ratio 1.14). Text and telephone reminders performed the same (risk ratio 0.99), with attendance of 67.8 per cent with no reminder, 78.6 per cent with a text and 80.3 per cent with a call. Text was 55 to 65 per cent cheaper per attendance than a phone call. If a receptionist is spending a morning ringing round, the evidence says a well-timed text does the same job for a fraction of the cost.

A 2026 NHS quality-improvement project in general surgery clinics at Mid and South Essex, published in BMJ Open Quality, shows what the sequence looks like in practice: DNAs fell from 12.2 per cent at baseline to 9.4 per cent with a telephone call 72 hours before, and to 8.1 per cent with automated texts five to seven days before and again the day before. The authors’ conclusion is worth keeping in mind: reminders cannot on their own resolve non-attendance without fixing contact data, booking capacity and two-way communication.

Results are as stated in each paper’s abstract. The BMJ Open, Cochrane and BMC papers are pooled reviews; the JGIM paper is a single-site US trial and the BMJ Open Quality paper is a single-trust NHS improvement project.

Put the studies together and a pattern falls out. One reminder is better than none; two or three are better than one; the message that recovers the most slots is the one sent while there is still time to move the appointment, not the one sent the night before. The NHS project above used five to seven days and one day. The confirmation at booking counts as the first message and costs nothing.

The second half of the pattern is what the message lets the patient do. If the only options are attend or cancel, a proportion of people who cannot make it will simply not turn up, because cancelling feels like admitting something. If the same text offers three alternative times that can be accepted in one tap, a share of would-be no-shows become moved appointments. The reminder is not the intervention. The reschedule link is.

The final piece is what happens to the released slot. A cancellation received three days out is only worth something if the slot is refilled, and it will only be refilled if the offer reaches a waiting list within minutes rather than after a receptionist has worked through a call sheet. The proportion of released slots refilled is the number almost no clinic measures, and it is the one that turns a cancellation policy from a deterrent into revenue.

NHS England has been piloting predictive no-show software since February 2023, when Mid and South Essex NHS Foundation Trust began using Deep Medical’s system to predict which appointments were likely to be missed, offer those patients more convenient times, and arrange intelligent back-up bookings so a slot was not wasted. In March 2024 NHS England reported that the trust had cut DNAs by 30 per cent over six months, preventing 377 missed appointments and allowing 1,910 additional patients to be seen, and announced a roll-out to ten further trusts. The same release cited University Hospitals Coventry and Warwickshire cutting DNAs from 10 per cent to 4 per cent in a pilot subset using process mining, and Sheffield Children’s sending 53,800 texts over twelve months with roughly 200 additional appointments attended a month.

Those are NHS England’s figures for NHS programmes, not vendor claims, but they are pilot results rather than controlled trials, and the comparison period is not published. The only randomised evidence on model-targeted reminders is a US primary-care trial across 5,840 appointments: adding a live reminder call for patients a random-forest model flagged at 15 per cent or more no-show risk cut the no-show rate from 36 per cent to 33 per cent overall, and from 42 per cent to 36 per cent among Black patients. Three points overall is a modest gain, bought with a phone call to a minority of bookings.

For a private clinic the practical lesson is narrower than the headlines. You do not need a predictive model to know which of your bookings are risky; your own booking history will tell you within a month. Targeting the live call at those, and sending texts to everyone else, is the cheap version of what the NHS is paying for.

Deposits work, and they cost bookings. Whether the trade is worth it depends on how full the diary is. What is not optional is the shape of the policy, because a clinic is a trader selling to consumers. The Competition and Markets Authority’s unfair contract terms guidance says a term that makes any substantial prepayment entirely non-refundable regardless of the circumstances "is more likely to be unfair". A genuine deposit may be kept in full only where it is a binding reservation, the patient is told up front that it is required and exactly when it would be lost, those circumstances are "clear and narrow", and it is "not normally more than a small percentage of the price". A sliding scale of cancellation charges is acceptable if it is prominent and never punitive.

That rules out the two policies clinics most often copy from each other: a large non-refundable deposit with no stated exceptions, and a full-fee charge for any cancellation whatever the notice. It rules in a small deposit, a stated notice period, and a scale that reflects what the clinic actually loses. The policy also has to be in the confirmation message, not only on the website.

For CQC-registered services there is a second constraint. The single assessment framework’s quality statement on equity in access expects providers to make sure "everyone can access the care, support and treatment they need when they need it", lists "waiting times, delays and cancellations" and "digital exclusion" among the subtopics it assesses, and expects reasonable adjustments and the removal of barriers. A reminder system that only works for people with smartphones, or a deposit that quietly excludes patients on low incomes, is the kind of thing an inspector now asks about. Appointment confirmations and reminders are service messages rather than marketing under the ICO’s guidance, so no marketing consent is needed to send them, but the contact data still has to be accurate and the patient still has to be able to reply.

Most of the mechanism above is a feature of practice-management software the clinic may already pay for. The table records what each vendor publishes, read in September 2026. Several of the best-known UK names publish no price at all, which is itself useful to know before a demo.

All prices as published on the vendor’s own page on the date shown; Cliniko and Twilio publish in US dollars. Where a vendor does not publish a price, the table says so rather than quoting a third-party estimate.

Take a hypothetical private clinic with 400 appointments a month, an average fee of £150, and a no-show rate of 8 per cent, which is the baseline NHS England reported for Mid and South Essex before its pilot. None of these are client figures; they are chosen to make the arithmetic visible. Change them for your own.

At 8 per cent, 32 appointments a month are missed. That is £4,800 a month in fees, £57,600 a year, before counting the clinician hours. If a reminder sequence performs as the BMJ Open pooled estimate suggests, a 25 per cent relative reduction, no-shows fall to 24 a month and eight fees are recovered: £1,200 a month, £14,400 a year. The messages cost almost nothing. Three texts per appointment is 1,200 messages a month; at Twilio’s published UK rate of $0.056 that is about $67, and at Cliniko’s 10 cents a credit it is $120. WriteUpp’s Solo plan includes 100 credits in £27.95 a month and sells top-ups on top.

The second gain is the refill. If the reschedule link turns some of the remaining 24 into cancellations with notice, and the waiting-list offer refills even one released slot in four, that is a further six appointments a month, £900, or £10,800 a year. Neither figure is guaranteed; both are the kind of number a clinic can measure within a quarter. Against that, an AI receptionist that answers the phone, sends the sequence and makes the targeted calls is priced at a level a 400-appointment clinic can test in a single quarter, which is the point of the next section.

Everything in this guide can be done with a practice-management system and a disciplined front desk. What an AI receptionist adds is the part the front desk cannot do at scale: answer every call including the ones that arrive while the receptionist is with a patient, send the reminder sequence on time without anyone remembering, make the targeted live call to the high-risk bookings, and offer a released slot to the waiting list within minutes rather than hours. The Cochrane evidence says a call and a text perform the same for the average patient, so the call should be reserved for the bookings where it earns its cost.

It does not fix wrong phone numbers, a diary with no capacity to reschedule into, or a policy that is unclear. The NHS project that cut DNAs from 12.2 to 8.1 per cent made exactly that point. It also does not remove the CQC duty to serve people who cannot use a smartphone, which is why a voice channel and a human fallback matter more in a clinic than in most businesses.

Augustova builds this as a fixed-price project. An AI adoption audit at £2,500 to £6,000, credited against a build, is usually enough to pull the last twelve months of bookings, find which appointment types and lead times carry the risk, and price the fix. A first agent, which for a clinic normally means the reminder sequence, the reschedule flow and the waiting-list refill, is £4,000 to £15,000; a custom AI system with the receptionist, targeted calling and integration into your practice software is £8,000 to £25,000. All fixed, excluding VAT, no day rates.

A clinic that changes three things at once will not know which one worked. Measure these before the change, then monthly.

This guide was first published in August 2026 and rebuilt on 14 September 2026. NHS activity figures are from NHS England Digital’s Hospital Outpatient Activity 2024-25 summary, published September 2025; the 2025-26 edition is due on 24 September 2026 and the figures will be updated when it appears. The reminder evidence is quoted from the abstracts of the papers named. NHS pilot results are NHS England’s own press releases, which report programme outcomes rather than controlled trials. Regulatory expectations are the CQC quality statement on equity in access, the CMA’s unfair contract terms guidance and the ICO’s direct marketing guidance, read at source. Software prices were read on each vendor’s own page in September 2026; where none is published the table says so. The worked example uses stated hypothetical figures and no client data.