Loading
Vendors promise a 50% cut in no-shows. The 2026 meta-analysis puts reminders at a risk ratio of 1.11, with telephone reminders clearing statistical significance and SMS falling short of it. We break down what the evidence supports, why targeting beat broadcasting in the two largest studies, what a missed appointment actually costs, and which widely quoted figures fail a source check.
Last updated: August 2026. Heading into 2026, medical practice leaders named no-shows their single biggest patient-access priority, ahead of online scheduling, phone access, and wait times (MGMA Stat, December 2025, n=236). Most advice on how to reduce no-shows stops at "send a reminder text." A 2026 meta-analysis of 10 controlled studies found something more useful: telephone reminders cleared the bar for statistical significance, and SMS reminders did not. Below is what the research actually supports, what moved the number in the two largest studies, what a missed appointment really costs, and which widely quoted figures do not survive a source check.
Key takeaways
- Reminders work, but the effect is smaller and far more variable than vendors advertise. Pooled across 10 controlled studies (8,236 participants), reminders raised attendance with a risk ratio of 1.11 (95% CI 1.05 to 1.19) in the Journal of Hospital Management and Health Policy, 2026.
- In that analysis, telephone reminders reached statistical significance (P=.002). SMS reminders did not (P=.07). The authors caution this was not a direct head-to-head test, and we flag our own commercial interest in that result below.
- The two largest documented drops came from targeting, not broadcasting. A UAE primary care network cut no-shows from 20.82% to 10.25% across 135,393 appointments by predicting who was at risk and having staff contact them (JMIR Formative Research, 2025).
- In a randomized trial, adding a live phone call for high-risk patients on top of the automated reminders everyone already received cut no-shows from 29.2% to 22.8% (Journal of General Internal Medicine, 2016).
- The "$150 billion a year" cost of missed appointments traces to an April 2017 trade-magazine article written by a scheduling vendor's marketing chief, not to independent research. We do not use it. A 2025 survey of 250 healthcare professionals puts the average practice loss at $22,872 a year.
There is no single national no-show rate. Published figures cluster between roughly 10% and 30%, and where you land depends far more on your patient mix and appointment type than on your reminder software. Two peer-reviewed anchors are useful here. A primary care network in the United Arab Emirates recorded a 20.82% no-show rate across the 67,429 appointments in its pre-intervention period (JMIR Formative Research, 2025). A randomized trial in US primary care recorded 29.2% in its control arm, though that arm was deliberately restricted to patients already flagged as high risk (Shah et al., Journal of General Internal Medicine, 2016).
On the patient side, 15% of US adults admitted to skipping an appointment without canceling in a 2025 survey of 750 Americans (Tebra, 2025). Before benchmarking yourself against anyone, measure your own rate for 90 days and split it by appointment type. A first-visit no-show and a recall no-show are different problems with different fixes.
For most practices, no. When MGMA polled 265 medical groups in August 2025, 60% said their no-show rate had stayed the same year over year and 13% said it had decreased. Only 27% reported an increase (MGMA Stat, August 2025). Roughly three in four practices are therefore holding steady or improving.
That makes the December 2025 finding more striking. No-shows still topped the 2026 patient-access priority list at 27%, with online scheduling at 24%, phone access at 22%, and wait times at 21% (MGMA Stat, December 2025). Practices are not reacting to a spike. They have worked out that the residual rate, the one that survives after they have already installed reminder software, is the expensive part. Leaders in the August poll who reported stable or falling rates credited consistent digital reminders, live outreach, easy cancellation and rescheduling, and in some cases deposits or penalty fees. Two of those five require someone to actually talk to the customer.
Yes, though the effect is more modest and more variable than most marketing suggests. The most current pooled evidence is a 2026 systematic review and meta-analysis by Al-Turbag, Mooney and Corry, covering 12 studies with 10 contributing 8,236 participants to the analysis (Journal of Hospital Management and Health Policy, Vol 10, March 2026). Reminders of any kind improved attendance with a risk ratio of 1.11 (95% CI 1.05 to 1.19).
Translate that into your schedule. Because attendance starts high, an 11% relative gain in attendance is a large relative cut in absences: a clinic at 75% attendance would expect to move to roughly 83%, turning a 25% no-show rate into something closer to 17%. Worth having, certainly. It is not the flat "50% reduction" promised on vendor pages, and the authors explicitly report "substantial heterogeneity across the included studies, indicating that reminder system effectiveness varies with contextual factors." Your mileage genuinely will vary.
In the 2026 meta-analysis, telephone reminders cleared the bar for statistical significance and SMS reminders did not. Telephone reminders, across 5 studies and 3,369 participants, produced a risk ratio of 1.11 (95% CI 1.04 to 1.19, P=.002). SMS reminders, across 4 studies and 4,636 participants, produced a risk ratio of 1.14 but with a confidence interval crossing 1.0 (95% CI 0.99 to 1.31, P=.07).
"Telephone reminders are associated with a significant improvement in attendance at Z=3.08 (P=0.002)... SMS reminders show a trend toward improved attendance, but the pooled effect did not reach conventional statistical significance."
Al-Turbag, Mooney and Corry, Journal of Hospital Management and Health Policy, March 2026
Not as far as we would like. We sell voice agents, so a result reading "calls beat texts" flatters us, and it deserves three caveats before anyone treats it as a headline. First, the same authors write that their "subgroup analysis explored the impact of reminder type but did not directly compare SMS and telephone reminders," and that context differences "limit direct comparisons." Second, this is a modest evidence base in a smaller journal, and larger reviews over the years have found SMS reminders effective. Treat it as the most recent pooled analysis on the question, which is still open. Third, a confidence interval crossing 1.0 means "not proven," which is a different thing from "does not work."
The defensible reading is narrower and still useful. A phone call has cleared an evidence bar that a text has not yet cleared, and the plausible mechanism is that a call is two-way. A text tells someone about an appointment. A call can discover that Tuesday no longer works and move them to Thursday before the slot is wasted.
Targeting. In both of the largest interventions in the literature, the win came from working out who was likely to miss and then having a real conversation with that smaller group. Sending more reminders to everyone was never the mechanism.
Emirates Health Services ran a random forest model with 86% accuracy over every scheduled appointment, surfaced the risk scores on a live dashboard, and had clinic coordinators proactively contact the high-risk bookings. Across 135,393 appointments, no-shows fell from 20.82% to 10.25%, an odds ratio of 0.43 (95% CI 0.42 to 0.45, P<.001).
Note the design honestly. This was a before-and-after study, so other changes over the period could have contributed, and the authors state that "the mitigation workflow is outside of EHRs, so we cannot quantify the efforts accurately." Vendor pages nonetheless cite this study as proof that "AI reminder calls cut no-shows by 50%." That is a misreading. The model decided who to contact; clinic coordinators ran the outreach, and the paper does not document the contact channel in detail. The transferable lesson is about targeting plus conversation, and it holds whether the caller is a coordinator or a voice agent.
Shah and colleagues took 2,247 primary care patients already flagged at above 15% no-show risk. Every patient in both arms kept receiving the standard automated calls. The intervention arm additionally got a live call from a coordinator seven days out who walked them through a concrete plan for getting there. No-shows fell from 29.2% to 22.8%, an absolute reduction of 6.4 points (95% CI -9.8 to -3.0, P<.01). That 6.4 points was earned on top of automation that was already running, which is the detail most summaries drop.
| Intervention | Study design | Measured effect |
|---|---|---|
| Any reminder vs none | Meta-analysis, 10 studies, 8,236 participants (JHMHP, 2026) | Attendance RR 1.11 (95% CI 1.05 to 1.19) |
| SMS reminder vs none | 4 studies, 4,636 participants (JHMHP, 2026) | RR 1.14 (95% CI 0.99 to 1.31), P=.07, not significant |
| Telephone reminder vs none | 5 studies, 3,369 participants (JHMHP, 2026) | RR 1.11 (95% CI 1.04 to 1.19), P=.002, significant |
| Live call added to existing automation, high-risk patients only | Randomized trial, 2,247 patients (JGIM, 2016) | No-shows 29.2% to 22.8%, 6.4-point absolute drop |
| Risk model + dashboard + coordinator outreach | Before-and-after, 135,393 appointments (JMIR, 2025) | No-shows 20.82% to 10.25%, OR 0.43 (95% CI 0.42 to 0.45) |
Start with a number that has a stated method. In a 2025 survey of 750 US adults and 250 healthcare professionals, practices reported an average of 14 missed appointments per month and an average annual loss of $22,872 (Tebra, 2025). That is a self-reported vendor survey, and we flag it as such. Its sample size and method are at least disclosed, which is more than most figures in this category can claim.
The better number is the one you calculate yourself, because it is the only one your bank will recognize. Take your average revenue per completed appointment, multiply by your monthly no-show count, then multiply by 12. Then add the second-order cost most owners forget: the slot could have been filled if you had known 48 hours earlier instead of finding out when nobody walked in. Learning about an absence in advance leaves you a rescheduling problem. Learning about it at the appointment time leaves you a dead hour.
You will see "missed appointments cost the US healthcare system $150 billion a year" repeated on hundreds of pages. It traces back to an April 2017 article in Health Management Technology by Jamie Gier, then chief marketing officer of the scheduling vendor SCI Solutions (citation via Medical Transportation Access Coalition). The arithmetic behind it is roughly a 30% national no-show rate multiplied by $200 per open 60-minute slot. That is vendor modeling published as trade commentary, and the inputs are nine years old. We treat it the way we treat every unsourced round number in this industry: useful as a signal that the problem is large, useless as evidence. When a page quotes it as a hard fact without naming its origin, apply the same caution to the rest of that page's numbers.
Fees and deposits are both common. Neither is free. MGMA found 42% of medical group leaders reported using a no-show fee, against 58% that did not (MGMA Stat, January 2025). Tebra's separate survey of healthcare professionals put it at 34%, with 16% of patients saying they had been charged one. The two surveys measure different populations, so treat the answer as a range. Deposits appear less often in the polling and work the same way: they shift risk onto the customer and leave the underlying cause in place.
The practical problem is that penalties address the wrong cause. Asked why no-shows were rising, practice leaders pointed to transportation, economic hardship, copay costs, and insurance limitations (MGMA Stat, August 2025). A penalty does not solve a transportation problem. It simply adds a bill to one. Fees are also off the table for Medicaid patients: providers who accept Medicaid agree to take the state's payment as payment in full, which is generally read to rule out billing the beneficiary for a missed appointment (42 CFR 447.15). State practice varies, so check your own program. In most cases it removes the option for exactly the population with the highest rates. In the 2025 patient survey, 33% of missed visits were attributed to work or scheduling conflicts and 14% to transportation. Almost all of that is solvable by rescheduling, and rescheduling requires someone to pick up the phone.
In several places. Pretending otherwise would cost you money. Someone who missed an appointment because of a serious diagnosis, a bereavement, or a financial crisis needs a person, and an automated follow-up call in that moment reads as tone-deaf. Complex clinical rescheduling that depends on lab timing, a referral window, or a specialist's availability usually needs staff judgment. High-value clients, whether a large commercial bid or a long-standing patient of record, often expect a named human to call. Any conversation heading toward a complaint should escalate straight to a person.
What people are genuinely bad at is volume and timing. Nobody on your front desk is going to call 60 customers at 7pm the night before, or answer the callback that arrives at 11pm from the patient who wants to move Tuesday to Thursday. That is the gap worth automating, and it is where the evidence points: the routine confirm-and-reschedule conversation, run consistently, at whatever hour suits the customer. Research on whether callers actually accept an AI on the line is covered separately in our review of the acceptance data.
Answering inbound calls and placing outbound ones are two different regulatory situations, and the distinction matters before you switch on reminder campaigns. In February 2024 the FCC confirmed that AI-generated voices count as an "artificial or prerecorded voice" under the TCPA (FCC Declaratory Ruling, February 2024). Those restrictions attach to calls that are made or initiated by the caller. In its follow-up rulemaking the Commission stated that the TCPA's requirements "do not extend to technologies used to answer inbound calls" (FCC 24-84, August 2024). Outbound reminder campaigns therefore generally depend on holding appropriate prior consent for that number and that purpose. Several states add their own AI-disclosure expectations, and state call-recording consent law applies in both directions. Our guide to AI receptionist disclosure laws covers the current picture. None of this is legal advice: confirm your position with your own counsel before you dial.
Healthcare adds a second layer. Where call content touches protected health information, the vendor is acting as a business associate, and HHS requires that relationship to be governed by a written business associate agreement (HHS, Business Associate Contracts). Ask any vendor for the executed document. A compliance badge on a marketing page is not one. Our guide on what to verify before you sign lists the specific questions worth putting to any provider, including us.
It closes the loop that reminder software leaves open. Where a text delivers a one-way notification, an AI voice agent calls, confirms, and if the answer is no, rebooks on the spot and updates your calendar and CRM. AIEmply answers 100% of calls in under 3 seconds, 24/7/365, handles unlimited simultaneous calls, speaks 50+ languages, and hands off warm to a human whenever a call needs one. That escalation path is what makes automating the routine majority safe. On the compliance questions above: AIEmply is HIPAA-ready for healthcare clients, GDPR and CCPA compliant, uses enterprise encryption, and does not share client data with third parties. It is the flagship product of Veltro Systems LLC.
Applied to the cycle, the work splits four ways. Confirm the appointment with a real conversation instead of a one-way blast. Capture the reschedule at the moment someone says they cannot make it, rather than letting them hang up and forget. Follow up on the ones who missed anyway, within hours instead of days. And answer the inbound call when the customer rings back, including on a Sunday evening. That last job matters because small businesses are poor at picking up in general: a 2016 industry survey of 85 small businesses found up to 62% of calls went unanswered (411 Locals, 2016).
| Stage | What usually happens | What a voice agent does |
|---|---|---|
| 48 hours out | Automated text goes out, no reply required | Calls, confirms, and asks whether the time still works |
| Customer cannot make it | They intend to call back and do not | Rebooks on the call, syncs calendar and CRM |
| They miss anyway | Front desk gets to it in a few days, if at all | Calls back the same day and offers the next slot |
| They ring back after hours | Voicemail, then a competitor | Answers in under 3 seconds, books the replacement |
Outbound follow-up campaigns start on the Growth AI Employee plan at $399/mo. The $149/mo Starter plan covers inbound answering and booking only, so if reminder and recovery calls are the goal, Growth is the entry point. This is a different failure point from the one covered in speed to lead, which is about how fast you respond to a brand new enquiry, and from the cost of missed calls, which is about leads that never reach you at all. A no-show is a lead you already won and then lost after the booking, which is why it stings more and why it deserves separate measurement.
One caveat first: every study cited above comes from clinical settings. Applying it to the trades and to professional services is reasoning by analogy, not evidence, and we would rather say so than imply a research base that does not exist.
With that stated, the mechanics carry over and the economics do not. A dental practice loses a fixed chair hour and a hygienist's time, and its recall list is usually long enough to make a same-day fill realistic given enough notice. An HVAC or plumbing company loses the truck roll and the technician's drive time, often costlier than the visit itself, and its version of a no-show is the customer who is not home inside the service window. A law firm loses a consultation slot booked weeks out, and a missed intake frequently means the case went elsewhere.
What travels across all three is the timing rule. The value of learning about a cancellation decays fast. Forty-eight hours out, you can refill the slot. Two hours out, you are usually eating it. Any process that shortens the gap between the customer knowing they cannot come and you knowing it is worth measuring against your own average revenue per appointment. Closing that gap takes a conversation.
Measure first, then target. Most practices skip straight to buying another reminder tool, which is exactly why their residual rate never moves.
Reminders help, by roughly 11% in attendance terms, with wide variation between settings. Telephone reminders have cleared an evidence bar that SMS has not, with the caveats above attached. And the largest documented reductions came from targeting the people most likely to miss and holding a two-way conversation with them. Broadcasting more notifications to everybody was never what moved the number. If you already send reminders and your no-show rate has not moved, more of the same will not fix it. Adding a conversation might.
AIEmply is more than a phone bot. It is a trained AI employee that confirms, reschedules, recovers, and answers the callback, then writes it all back to your calendar and CRM. 100% Answer Rate • Ready in 1–2 Weeks • Performance Guarantee. Billing starts only after your AI Employee is live. If the first month delivers no measurable result, the next month is free. No credit card required.
Compare plans, try a live demo, or book a 15-minute consultation and we will map your no-show cycle to a call flow before you commit to anything.
Published rates cluster between roughly 10% and 30%, so there is no universal target. A UAE primary care network measured 20.82% before intervening, and a US trial recorded 29.2% among patients already flagged as high risk. Rather than benchmarking against a national average, measure your own rate for 90 days, split it by appointment type, and treat any sustained improvement against that baseline as the goal.
Yes, but less dramatically than commonly claimed. A 2026 meta-analysis pooling 10 controlled studies and 8,236 participants found reminders improved attendance with a risk ratio of 1.11 (95% CI 1.05 to 1.19). For a clinic at 75% attendance that is a move to roughly 83%. The authors also reported substantial heterogeneity, meaning results vary considerably by setting and patient population.
In the 2026 meta-analysis, telephone reminders reached statistical significance (RR 1.11, P=.002) while SMS reminders did not (RR 1.14, P=.07). The authors caution that they did not run a direct head-to-head comparison, and larger reviews have found SMS effective. The defensible reading is that phone calls have cleared an evidence bar SMS has not in this analysis, likely because a call is two-way and can capture a reschedule.
A 2025 survey of 250 healthcare professionals reported an average of 14 missed appointments per month and an average annual loss of $22,872. Calculate your own figure by multiplying average revenue per completed appointment by monthly no-shows, then by 12. We do not use the widely quoted "$150 billion a year" national figure. It comes from an April 2017 trade-magazine article by a scheduling vendor's marketing chief, built on estimated inputs.
They are widely used, with 42% of medical group leaders reporting one in a January 2025 MGMA poll, but they target the wrong cause. Practice leaders cite transportation, economic hardship, and copay costs as leading drivers, and fees are generally read as off the table for Medicaid patients under 42 CFR 447.15, which removes the option for a high-rate population in most states. Making rescheduling effortless addresses more missed appointments than penalizing them does.
Yes. AIEmply places outbound confirmation, reminder, and recovery calls, rebooks on the call when a customer cannot attend, and syncs the change to your calendar and CRM. It answers inbound callbacks in under 3 seconds, 24/7/365, in 50+ languages, and hands off warm to a human when a call needs one. Outbound follow-up campaigns start on the Growth AI Employee plan at $399/mo.
Generally yes for outbound calls, and the rules differ from inbound answering. Inbound call answering sits outside the reach of the TCPA, while outbound automated calling to consumers usually depends on holding appropriate prior consent for that number and purpose. Several states add AI-disclosure expectations, and healthcare calls involving protected health information require a signed business associate agreement. Confirm your specific position with your own legal counsel.
No. Automate the routine confirm-and-reschedule conversation, which is high volume and time-sensitive. Keep humans on emotionally sensitive situations, complex clinical rescheduling that depends on labs or referrals, high-value clients who expect a named contact, and any call trending toward a complaint. A warm handoff path is what makes automating the routine majority safe.