Appointment Reminders That Actually Reduce No-Shows

Start with the size of the effect, not the promise

Vendor pages in this category quote relative reductions, usually with an "up to" in front of them: cut no-shows by 38%, by 80%, by half. Those figures are almost never wrong so much as unanchored, because a relative reduction is meaningless without the base rate it applies to. A 30% relative reduction on a 10% no-show rate is three appointments per hundred. The same 30% on a 4% rate is barely more than one. Before you evaluate any reminder product, work out what a few points are worth in your schedule, because that is the size of the prize.

The Cochrane systematic review of mobile phone messaging reminders, updated by Gurol-Urganci and colleagues in 2013, pooled seven studies covering 5,841 participants and found that text message reminders improved attendance compared with no reminder at a risk ratio of 1.14, with a 95% confidence interval of 1.03 to 1.26, which the reviewers rated moderate-quality evidence. That is a real effect with a confidence interval that does not cross one, and it is also, in plain terms, roughly a tenth more people showing up, not a transformation.

The broader systematic review by Hasvold and Wootton, published in the Journal of Telemedicine and Telecare in 2011 and covering 29 studies of telephone, SMS, and automated call reminders, reported a relative reduction in non-attendance of roughly 29% for automated reminders and about 39% for manual telephone reminders. Again, relative. Applied to a clinic already running at a 12% no-show rate, the automated figure is about three and a half points.

And the ceiling is real. A 2016 analysis by Kheirkhah and colleagues in BMC Health Services Research, covering more than a decade of outpatient scheduling data at a large VA medical center, found a mean no-show rate of 18.8% and an estimated cost of $196 per no-show. Adding a centralized phone reminder call to that system moved the no-show rate from 16.3% to 15.8%. Half a percentage point, from an intervention that on paper sounds identical to the ones that produce bigger numbers elsewhere. The difference is almost always in the details below, not in the technology.

Two reminders beat one, and the spacing is the variable

The single most actionable result in this literature is that reminder count and timing are not matters of taste, and the trials that tested them are large.

Steiner and colleagues ran a three-armed randomized trial at Kaiser Permanente Colorado, published in the American Journal of Managed Care in 2018, comparing a single reminder three days before a primary care visit, a single reminder one day before, and both. Across 54,066 randomized patients, the missed-appointment rate was 4.4% for patients who got both reminders, against 5.8% for the three-day-only arm and 5.3% for the one-day-only arm. The benefit was concentrated in patients a clinical prediction rule had flagged as high risk of missing, which is a useful hint about where to spend a second message if you are rationing them.

A later pragmatic randomized study at Kaiser Permanente Washington, published in The Permanente Journal in 2022, tested one text reminder against two across 125,076 primary care visits and 33,593 mental health visits. The second text reduced the chance of a no-show by 7% in primary care, a risk ratio of 0.93 with a 95% confidence interval of 0.89 to 0.96, and by 11% in mental health, a risk ratio of 0.89 with an interval of 0.86 to 0.93. It also cut same-day cancellations in primary care by 6%, though it did not move same-day cancellations in mental health at all.

The realist systematic review by McLean and colleagues, published in Patient Preference and Adherence in 2016, arrived at a compatible recommendation from a different direction: send reminders a minimum of two to three days in advance of the appointment.

The mechanism behind all three is worth stating because it tells you what to build. The far reminder and the near reminder are doing different jobs. The far one exists to open a cancellation window, so that someone who already knows they cannot make it tells you while the slot is still refillable. The near one exists to catch the ordinary drift of a busy day. A single reminder has to do both jobs and does neither particularly well: too early and the day-of memory is gone, too late and the person who cancels has handed you a slot you cannot fill.

The wording moves the number more than the channel does

The most striking result in this field did not involve buying anything. Hallsworth and colleagues, publishing in PLOS ONE in 2015, ran two randomized controlled trials in UK hospital outpatient clinics, covering 10,111 appointments in the first and 9,848 in the second. Appointments were randomly assigned to one of four SMS reminder messages, all issued five days in advance. The message that stated the cost of a missed appointment to the health system produced a did-not-attend rate of 8.4%, against 11.1% for the hospital's existing message. The second trial replicated the effect at 8.2%.

That is close to a three-point absolute improvement from editing a text message. No new platform, no change in timing, no additional send. It is also the clearest evidence available that the message body is a design surface rather than boilerplate, and most reminder systems ship with boilerplate.

Two caveats before anyone copies it literally. The trials ran in a public health system where the cost of a wasted slot lands on a service the recipient funds through taxes, and that framing does not transfer cleanly to a private business telling a customer what their absence costs the business. The McLean review's finding is the more portable one: reminders that carry information beyond the bare fact of the appointment, which the review calls "reminder plus," may outperform simple reminders, and reminders should be framed to ask people to cancel and reschedule appointments they no longer want.

In practice that means a reminder should carry the business name, the date and time or arrival window, what the appointment is for, anything the person needs to bring or do beforehand, and exactly one obvious way to change it. What it should not carry is a promotion, for reasons covered in the consent section below.

Channel: text, call, or email

The Cochrane review also compared text reminders against telephone call reminders directly, across three studies and 2,509 participants, and found a similar impact on attendance, rated moderate-quality. The cost finding is the one that decides most deployments: in two of the included studies, the relative cost of text reminders per attended appointment was 55% to 65% of the cost of telephone reminders.

Hasvold and Wootton's larger review puts a thumb on the other side of the scale, with manual telephone reminders reducing non-attendance by about 39% against roughly 29% for automated ones. These are not contradictory findings. A person calling a person reaches people a text does not reach, handles a "can I move it" in the same breath, and costs staff time that scales with volume. That is precisely the trade a business has to make deliberately rather than by default.

• Text as the default channel. It is the cheapest per attended appointment, it is asynchronous, and it leaves a record the recipient can look at again.
• Calls as an escalation, not a baseline. Reserve them for high-value slots, for people with a history of missing, and for anyone who has not responded to two texts.
• Email as a supplement. It is fine for the far reminder where there is detail to convey, and unreliable as the last touch before an appointment.
• Whatever you choose, watch who it fails. McLean's review recommends monitoring whether any specific group of patients is being disadvantaged by the chosen reminder system, which in a service business translates to checking whether your text-only programme quietly has a worse attendance rate among the customers least likely to be carrying a smartphone.

A reminder nobody can act on is only a notification

The reminder's job is not to inform. It is to produce one of three outcomes: the person shows up, the person tells you they cannot, or the person moves the appointment. Two of the three require a reply path that goes somewhere, and this is where most reminder setups quietly fail.

If the reminder comes from a number that cannot receive replies, you have built a system that can only prevent the forgetting kind of no-show and is structurally incapable of converting the other kind into a cancellation you can refill. Worse is a number that accepts replies into an inbox nobody watches: the customer told you, in writing, that they were not coming, and you held the slot anyway. That failure mode is invisible until you go looking for it, so go looking for it. Send yourself a cancellation reply and see what happens to it.

What a working reply path looks like is unglamorous. A reply that says cancel actually releases the slot in the calendar of record. A reply that says reschedule either offers real open times or hands off to a person with the context attached. A reply that says something else entirely, which is most of them, reaches a human. The booking mechanics are covered in how appointment automation works, and the failure modes that make these systems get switched off are catalogued in the common mistakes guide.

One warning about the evidence here. Search for two-way reminders and you will find confident numbers, a 62% reduction here, an 8 to 12% additional reduction there. Every one we could trace is a vendor case study: self-reported, uncontrolled, single-site, and published by the company selling the feature. The mechanism is sound and the reasoning above stands on its own, but those figures are marketing, not evidence, and it is worth knowing which is which before you repeat one in a budget meeting.

Reminders fix forgetting, and forgetting is only part of the problem

Parsons, Bryce, and Atherton published a systematic review in the British Journal of General Practice in 2021 on which patients miss primary care appointments and why. The reasons that recur across the included studies are forgetting the appointment, difficulty cancelling it, the appointment time being inconvenient, being too ill to attend, and no longer needing the appointment because the problem resolved. The review also found that younger patients, those of lower socio-demographic status, and those from minority ethnic groups were more likely to miss appointments.

Only the first of those is a reminder problem. Read the rest as a work list, because each one has a fix that is not a reminder:

• Difficulty cancelling is a process problem, and it is the one a reminder system is best placed to solve, provided the reply path above actually works. Someone who tried to cancel and could not is a no-show you manufactured.
• An inconvenient time is a scheduling problem. If your no-shows cluster in the first slot of the morning or the last of the day, no amount of messaging will fix what the calendar is doing.
• No longer needing the appointment is a demand problem, common in anything with a long booking lead time, and the reminder's only useful role is to surface it early enough to refill the slot.

There is a further limit worth internalizing. The McLean review notes that reminder effects are short-term and do not produce persistent behaviour change. A reminder programme is an ongoing operating cost that buys an ongoing attendance improvement. Switch it off and the rate goes back. Nobody selling one is likely to phrase it that way.

The slot you refill is worth more than the no-show you prevent

For any business with fixed capacity, a chair, a bay, a technician's day, an hour of a professional's time, the arithmetic is unforgiving: an empty slot is not deferred revenue, it is revenue that expired at the appointment time. The Kheirkhah analysis put the cost of a single outpatient no-show at $196 in that VA setting, which is a useful order of magnitude even though your own number will differ.

This is why the cancellation rate is the metric most reminder buyers ignore and should not. A reminder programme that converts would-be no-shows into cancellations 48 hours out looks, on a naive dashboard, like it increased your cancellations. What it actually did was turn dead slots into refillable ones. The dental case is the clearest illustration, because dental scheduling combines fixed chair capacity with a high rate of short-notice cancellation; the numbers and the backfill mechanics are in appointment automation for dental practices.

The corollary is that a reminder system without a waitlist or a backfill path is only doing half the job. Knowing about the gap early is worth very little if nothing happens with the knowledge. Whoever owns the schedule needs the cancellation to reach them in time to act, and ideally the system offers the freed slot to the next person automatically.

Almost all of this evidence is from healthcare, and that matters

Every trial cited above ran in a clinic, a hospital outpatient department, or a primary care practice. That is not an accident of our searching: healthcare systems have the scheduling data, the institutional research capacity, and the financial incentive to run randomized trials on attendance. Dental practices, HVAC companies, cleaning services, salons, and real estate teams have none of those things, and the result is that the non-healthcare side of this question has no comparable trial base at all. What circulates instead is vendor-reported case studies and round numbers with no methodology attached.

So be explicit about what transfers and what does not. The mechanism transfers: forgetting is not a medical phenomenon, and neither is the friction of trying to cancel something. The recommendation to send two reminders, spaced days apart, with a working reply path, rests on a mechanism that is the same for a boiler service as for a clinic appointment.

The numbers do not transfer. Your base no-show rate is different, your appointment lead times are different, the consequence of missing is different, and the cost-message effect from the Hallsworth trials depends on a public-funding framing that does not exist in a private transaction. Anyone quoting you a percentage improvement for your industry is extrapolating, whether or not they say so.

Which leaves one honest conclusion: adopt the design principles from the trials, then measure the effect on your own schedule rather than believing a transferred number.

How to measure whether your reminders are working

Most businesses that install reminders never find out whether they helped, because the measurement is set up in a way that cannot answer the question. Four things make the difference.

• Define the metric before you change anything. A no-show is an appointment that was still on the books at its start time and nobody arrived. An advance cancellation is a different event, and the two move independently, sometimes in opposite directions. Dental data from roughly 3,400 practices, discussed in the dental appointment automation guide, shows the two rates sitting at very different levels in the same practices.
• Use the right denominator. No-show rate is missed appointments divided by scheduled appointments in the period, not divided by appointments that were kept. Dividing by kept appointments inflates the rate and makes any improvement look larger than it is.
• Hold something back if you can. The cleanest evidence you will ever get about your own business is a holdout: alternate reminders on and off by appointment ID, or by day of week, for a few weeks. It costs a small number of reminders and it is the only way to separate the reminder's effect from a seasonal swing.
• Respect the sample size. At a 10% base rate, detecting a two-point improvement with any confidence takes thousands of appointments. A single practice comparing one month to the next is looking at noise most of the time. Compare quarters, expect the number to bounce, and do not rip out a working system because a bad fortnight landed in the middle of it.

One reporting note that saves arguments later: track prevented no-shows and recovered slots separately. The first is the reminder doing its classic job. The second is the cancellation window doing its job, and it is usually the larger number.

Consent, frequency, and opt-out

None of this is legal advice, and the rules turn on what kind of business you are and where your customers are. But appointment reminders are outbound messages, and outbound is a different legal regime from answering calls people place to you. The general framework, including the FCC's 2024 ruling that AI-generated voices count as artificial voices under the Telephone Consumer Protection Act, is in the TCPA guide.

Healthcare providers have a narrow exemption and should know its edges precisely. Under the FCC's 2015 omnibus TCPA order, certain healthcare messages sent by or on behalf of a HIPAA covered entity or business associate can go to a wireless number without prior express consent, codified in 47 C.F.R. § 64.1200, with the wireless provision at paragraph (a)(9)(iv) and the parallel residential-line provision at (a)(3)(v). The conditions are strict and cumulative: the message must fall within the listed treatment-related categories, which include appointment and exam reminders, pre-operative instructions, lab results, prescription notifications, and post-discharge follow-up; it must be free to the end user; it must not contain telemarketing, advertising, billing, or debt-collection content; it is capped at one message per day and three per week per patient; voice messages must run a minute or less and texts 160 characters or less; and every message must carry an opt-out that is honored. Attach a promotion to the reminder and the message leaves the exemption entirely, at which point it is telemarketing and needs prior express written consent.

Outside healthcare, the practical position is simpler but not automatic. A reminder about an appointment the customer themselves booked is informational rather than promotional, and the defensible practice is to capture consent at the point of booking, keep the message purely about the appointment, include an opt-out, and keep marketing out of it. Bundling a seasonal offer into a service reminder is the single most common way a compliant programme becomes a non-compliant one.

Revocation deserves its own line in your process. Since April 11, 2025, a consumer's revocation of consent made by any reasonable method must be honored, and processed within ten business days. That means a STOP reply, or a spoken request on a call, has to propagate across voice and SMS together rather than stopping one channel while the other keeps sending.

If reminder content includes anything about a patient's care, the HIPAA questions stack on top of the TCPA ones, particularly around how much detail belongs in a message that may appear on a lock screen. Those are covered in HIPAA and AI voice agents. Confirm the specifics for your business and your states with your own counsel; both the consent rules and the state-level picture have been actively changing.

A default schedule worth starting from

Nothing below is a guaranteed configuration, and the point of the measurement section is that you should expect to change it. But this is where the evidence points, and it is a better starting position than whatever a platform ships with.

• Booking confirmation, immediately. Not a reminder, but it is where consent is captured and where the person learns that replying works.
• First reminder, three days out. Far enough ahead to be actionable, consistent with the McLean review's two-to-three-day minimum and with the far arm of the Steiner trial. Its job is to invite a cancellation or a reschedule, so the wording should say so plainly.
• Second reminder, one day out. This is the pairing that outperformed either single reminder in the Steiner trial. Keep it short and make the action obvious.
• For arrival windows rather than fixed times, add a day-of message when the technician is dispatched. Field service is a different shape of problem: the customer needs to be present, not to travel, so the near-term message is about narrowing the window rather than jogging a memory.
• Escalate to a call for the appointments that justify it: high-value slots, first appointments with new customers, and anyone with a history of missing. The evidence says human calls work slightly better and cost more, so spend them where the slot is worth it.
• Stop at the caps. Healthcare senders are bound by one message a day and three a week. Everyone else should treat that as a sensible ceiling anyway, because the fastest way to lose a channel is to make people opt out of it.
• Instrument all of it from day one. Reminder sent, reply received, cancellation processed, slot refilled. If you cannot see those four numbers, you cannot tell which part is working.

Sources

• Gurol-Urganci I, de Jongh T, Vodopivec-Jamsek V, Atun R, Car J, "Mobile phone messaging reminders for attendance at healthcare appointments," Cochrane Database of Systematic Reviews (2013), CD007458.pub3 — supports the risk ratio of 1.14 (95% CI 1.03 to 1.26) from seven studies and 5,841 participants, the finding of similar impact to telephone reminders from three studies and 2,509 participants, and the relative cost of text reminders at 55% to 65% of telephone reminders per attended appointment.
• Hasvold PE, Wootton R, "Use of telephone and SMS reminders to improve attendance at hospital appointments: a systematic review," Journal of Telemedicine and Telecare 17 (2011), 358-364 — supports the review of 29 studies and the approximate 29% and 39% relative reductions in non-attendance for automated and manual telephone reminders.
• Steiner JF et al., "Optimizing Number and Timing of Appointment Reminders: A Randomized Trial," American Journal of Managed Care 24(8) (2018), 377-384 — supports the three-arm design, 54,066 randomized patients, the 4.4%, 5.8%, and 5.3% missed-appointment rates for the combined, three-day-only, and one-day-only arms, and the larger effect in patients identified as high risk.
• "Pragmatic Randomized Study of Targeted Text Message Reminders to Reduce Missed Clinic Visits," The Permanente Journal (2022), doi 10.7812/TPP/21.078 — supports the 125,076 primary care and 33,593 mental health visits, the no-show risk ratios of 0.93 (95% CI 0.89 to 0.96) and 0.89 (0.86 to 0.93), and the same-day cancellation findings.
• Hallsworth M et al., "Stating Appointment Costs in SMS Reminders Reduces Missed Hospital Appointments: Findings from Two Randomised Controlled Trials," PLOS ONE 10(9) (2015), e0137306 — supports the two trials of 10,111 and 9,848 appointments, the four message variants issued five days ahead, and the did-not-attend rates of 8.4% against 11.1%, replicated at 8.2%.
• McLean SM, Booth A, Gee M, Salway S, Cobb M, Bhanbhro S, Nancarrow SA, "Appointment reminder systems are effective but not optimal: results of a systematic review and evidence synthesis employing realist principles," Patient Preference and Adherence 10 (2016), 479-499 — supports the two-to-three-day minimum lead time, the "reminder plus" finding, the recommendation to frame reminders around cancelling and rescheduling, the recommendation to monitor for disadvantaged groups, and the point that reminder effects are short-term rather than persistent behaviour change.
• Parsons J, Bryce C, Atherton H, "Which patients miss appointments with general practice and the reasons why: a systematic review," British Journal of General Practice 71(707) (2021), e406 — supports the list of reasons for missed appointments and the patient characteristics associated with missing them.
• Kheirkhah P, Feng Q, Travis LM, Tavakoli-Tabasi S, Sharafkhaneh A, "Prevalence, predictors and economic consequences of no-shows," BMC Health Services Research (2016) — supports the 18.8% mean no-show rate, the $196 average cost per no-show, and the move from 16.3% to 15.8% after adding a centralized phone reminder, in a VA medical center's outpatient clinics.
• FCC 2015 Omnibus TCPA Declaratory Ruling and Order (FCC 15-72) and 47 C.F.R. § 64.1200, wireless healthcare provision at (a)(9)(iv) and residential at (a)(3)(v) — support the healthcare exemption, the permitted message categories, the free-to-end-user requirement, the one-per-day and three-per-week caps, the one-minute and 160-character limits, and the opt-out requirement.
• FCC consent-revocation rules effective April 11, 2025 — support the requirement to honor revocation made by any reasonable method and to process it within ten business days.
• Figures attributed above to vendor case studies, including the two-way reminder reduction claims, are identified as such deliberately: they are self-reported and uncontrolled, and are cited as examples of marketing claims rather than as evidence.
• None of the above is legal advice. Consent, revocation, recording, and health-information rules vary by state and industry and are actively changing; confirm the specifics with your own counsel before a reminder programme goes live.