How Long Will Customers Wait Before Leaving?
No published study measures how long a shop, café or salon customer waits before walking out. Leaving gets counted in hospitals and on phone lines, and almost nowhere else. So the honest answer has to be built from those places, and what they measure turns out not to be the same behaviour.
Where leaving is counted, the published rates sit a long way apart. They run from 0.15% of registered patients at a children’s hospital in Tokyo, on data from 2014 to 2017 (Cureus, 2024). The top of the range is 12.12% at a hospital near Naples, on data from 2019 to 2023 (BMC Emergency Medicine, 2025).
There’s no threshold in the data either. Across three emergency departments in the United States, every 20-minute increase in the median stay of discharged patients came with two more patients leaving per day (West JEM, Providence, 2025). That’s a slope, not a cliff edge.
And one of the predictors that shows up isn’t the clock at all. It’s how many people are already waiting when someone arrives.
Key Takeaways
- Nobody has published a measurement of how long a shop, café or salon customer waits before walking out. The figures in circulation come from somewhere else.
- Where leaving is counted, in hospitals, the rates vary enormously. They run from 0.15% of registered patients at a Tokyo children’s hospital, on 2014 to 2017 data (Cureus, 2024). The top figure is 12.12% near Naples, on 2019 to 2023 data (BMC Emerg Med, 2025). Those services are not all counting the same thing.
- Leaving rises with the wait rather than switching on at a point. In three United States emergency departments, every extra 20 minutes in the median stay of discharged patients came with two more people leaving per day (West JEM, Providence, 2025).
- What a person can see counts as well as how long they wait. At one Canadian hospital, each additional five people already waiting raised a new arrival’s adjusted odds of leaving by 16.9% (West JEM, Ottawa, 2025).
Almost nobody counts how long customers wait before leaving
The measurement doesn’t exist for shops, cafés or salons, and managing a salon or clinic waiting room has to be done without it. Health services are the main exception: an emergency department records who arrived and who was seen. Even so, two neighbouring jurisdictions count different things.
Start with England. NHS England’s supplementary analysis of the Emergency Care Data Set reports attendances, admissions and twelve-hour waits, broken down by patient demographics (NHS England). None of it counts a patient who left. That is a statement about that one workbook, not about every English publication.
Northern Ireland does count something, and it isn’t the same thing. It publishes the share of attendances that “left before their treatment was complete” (Department of Health, 2026). That is a wider category than leaving without being seen. The department warns readers to be “cautious when making direct comparisons between Northern Ireland and other UK Jurisdictions as waiting times may not be measured in a comparable manner”.
The mismatch runs through the whole evidence base. Four things the published measurements disagree about:
- What counts as leaving. Northern Ireland counts attendances that left before treatment was complete (Department of Health, 2026). The North American and Italian studies count patients never seen at all (West JEM, Ottawa, 2025; BMC Emerg Med, 2025).
- When the clock starts. The Italian work measures from registration to the first medical evaluation. One paper states plainly that this is not the total length of stay in the department (BMC Emerg Med, 2025).
- Who is in the denominator. The Ottawa study counts patients who left before triage separately: 2,716 of 170,536 visits, against 15,473 who left after triage (West JEM, Ottawa, 2025). Two Turkish studies count adults only (Medicine (Baltimore), 2024; Postgraduate Medicine, 2026).
- Which sites are excluded. Northern Ireland’s June 2026 figure leaves out two urgent care centres, Craigavon and Daisy Hill, because their data sits on another system (Department of Health, 2026).
So what would a usable café figure need? Someone would have to record the moment a person joined the queue, and the moment they left. Nobody registers for that queue, and nobody has to stand in it. That belongs to taking the measurement yourself.
Where it is counted, what do the numbers say?
Nine measurements published between 2024 and 2026 put the rate between 0.15% and 12.12% (Cureus, 2024; BMC Emerg Med, 2025). Not all of them count it the same way, or over the same span. That range is evidence about how differently the counting is done, not an estimate of anything.
| Setting and country | What was counted | The figure | Data period |
|---|---|---|---|
| Children’s hospital, Tokyo, Japan | Left without being seen | 168 of 112,059 registered patients, 0.15% (Cureus, 2024) | April 2014 to March 2017 |
| Tertiary university hospital, Türkiye | Left without being seen. Adults 18 and over only | 806 of 188,313 visits, 0.43% (Medicine (Baltimore), 2024) | June 2021 to June 2023 |
| Taksim Training and Research Hospital, Türkiye | Left without being seen. Adults only | 9,737 of 1,058,817 visits, 1.0% (Postgraduate Medicine, 2026) | January 2020 to June 2025 |
| 307 hospitals, California, United States | Left without being seen, as reported by the hospitals | 3,093,129 of 119,651,721 visits, 2.6% (JAMA Network Open, 2026) | 2015 to 2024 |
| University Hospital of Salerno, Italy | Left without being seen | 39,188 of 688,870, 5.68% (Scientific Reports, 2024) | 2014 to 2021 |
| Three emergency departments, Providence, Rhode Island, United States | Left without being seen. Urban hospitals; the authors say these rates are not representative of the national median | Median rate 6.0% across 373,388 visits (West JEM, Providence, 2025) | October 2022 to June 2024 |
| Emergency departments, Northern Ireland | Left before their treatment was complete, a wider category than leaving without being seen. Excludes two urgent care centres | 6.8% of attendances (Department of Health, 2026) | June 2026 |
| The Ottawa Hospital, Canada | Left without being seen. Those who left before triage are counted separately | 15,473 of 170,536 visits, plus 2,716 pre-triage (West JEM, Ottawa, 2025) | May 2022 to April 2024 |
| Maresca Hospital, Torre del Greco, Italy | Left without being seen | 9,774 of 80,614 patients, 12.12% (BMC Emerg Med, 2025) | 2019 to 2023 |
These are not nine estimates of one thing. They are counts of different behaviours, in different health systems, over different years, on different bases. An average across them would mean nothing.
The clearest evidence that the rate belongs to conditions rather than customers comes from one site. At the hospital near Naples, the yearly share leaving without being seen ranged from 7.57% in 2019 to 16.14% in 2022 (BMC Emerg Med, 2025). Same hospital, same catchment, more than double in three years.
The largest dataset available moves the same way. Across 307 California hospitals, the rate swung between 2.2% and 3.4% from 2015 to 2024. At the 95th percentile of hospitals it reached 9.9% in 2022 (JAMA Network Open, 2026).
Is there a point where people give up?
No published evidence points to one. Across three emergency departments in one Providence, Rhode Island medical system, longer stays for discharged patients came with more people leaving. Two more patients left per day for every 20-minute increase in the median stay of discharged patients (West JEM, Providence, 2025). The model fitted a straight line, so it could not have found a tipping point even if one existed. What it does show is a steady climb.
Check the scope before carrying it anywhere. All three sites were urban: a teaching adult unit, a community one and a children’s one. The authors say their baseline rates were not representative of the national median (West JEM, Providence, 2025).
The second piece of evidence is quieter and may be the most useful thing here. The paper records a mean “waiting time for take-over” of 210.0 minutes for the group who left, against 124.3 minutes for the group who stayed (BMC Emerg Med, 2025). It defines that term as the time between registration and the first medical evaluation. That is not the total length of stay in the department, and the paper says so plainly.
Now look at the spread around those two means. In the same table, the standard deviation is 265.4 minutes for the group who left. For the group who stayed it is 202.1 minutes (BMC Emerg Med, 2025). In both groups the spread is wider than the average it sits around. A mean shaped like that can’t work as a threshold, even in principle.
A much larger Italian study points the same way. It covered 688,870 patients registered between 2014 and 2021 at the University Hospital of Salerno. The factor that most influenced the abandonment rate was the waiting time for take-over (Scientific Reports, 2024). Waiting matters, but it doesn’t switch anything on at a fixed point.
None of these figures says anything about how the wait felt to the person in it. That is a separate quantity, covered in why waiting feels longer than it is.
What an arriving customer can see changes the odds
One predictor here isn’t a duration at all. It’s a headcount. At The Ottawa Hospital, researchers analysed 170,536 emergency visits between 2022 and 2024. Each additional five patients already waiting to be seen raised a new arrival’s adjusted odds of leaving by 16.9% (West JEM, Ottawa, 2025).
Adjusted odds, per five people, measured at the moment of arrival. Not the wait itself.
A customer can’t see the clock running on the order ahead of them. They can see how many people stand between them and the counter.
In their discussion the authors offer a hypothesis, not a result. Patients may be more tolerant when they can see movement, and less tolerant in a static waiting room (West JEM, Ottawa, 2025).
Aksin and colleagues tested queues that empty fast early and then slow, matching a steady queue’s total wait. The early burst held people longer (Production and Operations Management, 2025). These were laboratory and online experiments, not real queues, and the abstract gives no sample sizes and no country.
How the wait is described changes things too, at identical durations. Reneging is the term for leaving a queue after joining it (International Journal of Tourism Research, 2025).
The same paper ran two restaurant waiting scenarios. Reneging was more frequent when the wait was described coarsely than when the same duration was described in finer grain (Int J Tourism Res, 2025). The authors propose anxiety as the mechanism, and report it for the interaction with a customer’s mindset rather than for wording alone. Those were made-up scenarios with 142 and 170 people in the United States. They record stated intentions, not observed behaviour.
One more detail matters. The Ottawa authors suggest moving patients between zones to create a sense of forward momentum. They argue it could cut leaving without changing actual wait times (West JEM, Ottawa, 2025). The citation behind it is a 1985 book chapter by David Maister. So a 2025 clinical paper’s advice rests on a management chapter forty years older.
What an arriving customer sees is partly a layout decision, argued out in single queue or multiple lines.
Who gives up first?
The Ottawa data answers this best, because it splits the rate by how urgent each patient’s condition was. The probability of leaving without being seen ran from 1.5% for the highest-acuity patients to 13.5% for the lowest (West JEM, Ottawa, 2025). How badly someone needs the thing they’re queueing for moves the answer by a factor of nine.
| Group of patients | Probability of leaving without being seen |
|---|---|
| CTAS 1, highest acuity | 1.5% |
| CTAS 2 | 7.0% |
| CTAS 3 | 8.7% |
| CTAS 4 | 12.2% |
| CTAS 5, lowest acuity | 13.5% |
| Arrived in the day, 7am to 5pm | 5.4% |
| Arrived in the evening, 5pm to midnight | 14.4% |
| Arrived at night, midnight to 7am | 13.9% |
Source: West JEM, Ottawa, 2025.
One presenting complaint is worth naming, because it runs against the grain. Headache carried the highest probability of leaving in that dataset, at 14.2% (West JEM, Ottawa, 2025). Vomiting or nausea followed at 11.4%, and chest pain with cardiac features sat at 9.1%.
Time of arrival moves it almost as much. At the same hospital, 5.4% of daytime arrivals left without being seen, against 14.4% of evening ones (West JEM, Ottawa, 2025).
The study records who left and under what conditions. It does not record why, and it did not ask.
Two limits decide how far this travels. The authors flag it as a single-centre study whose findings may not generalise. They also note their department has no internal waiting room, so departments with one may see different patterns. Beyond that, nothing here separates patience from crowding, triage or the definition in use. So these figures can’t be stacked into a ranking of settings.
The one queue Britain does measure is on the phone
Britain publishes waiting data for a queue nobody can see. Between April and June 2026, callers to HMRC waited on average 11 minutes and 35 seconds. Of those callers, 87.9% got through to an adviser (GOV.UK, 2026).
Be careful with that second figure. HMRC doesn’t separate people who hung up from calls that were deflected or cut off. So the remainder isn’t a walkaway rate, and it shouldn’t be used as one.
Ofcom measures the same behaviour in telecoms. In 2024 the two halves of the market moved in opposite directions: mobile waits fell and fixed-line waits rose (Ofcom, 2025).
| Publisher and period | What was measured | The figure |
|---|---|---|
| HMRC, April to June 2026 | Average wait for a call to be answered | 11 minutes 35 seconds |
| Ofcom, 2024 | Average time for mobile customers to reach an agent | 1 minute 52 seconds, down from 2 minutes 24 seconds in 2023 |
| Ofcom, 2024 | Industry average call waiting time, broadband and landline | 2min 1s, up 13 seconds on the 1min 48s average in 2023 |
Sources: GOV.UK, 2026; Ofcom, 2025.
One term needs care here. Ofcom uses “abandoned call” in two unrelated senses. In its consumer guidance, an abandoned call is one where the phone rings and then ends when you pick up, placed by an organisation calling you (Ofcom). In its customer service reporting, the same regulator analyses call abandonment by people waiting to be answered (Ofcom, 2025). Same phrase, two different behaviours.
For queues in shops, UK public data holds nothing equivalent. No national statistics series, no regulator return, no trade body count. A queue outside a bakery isn’t a reportable quantity, so nothing reports it.
The tolerance figures that circulate for shop queues were measured on telephone lines. The figure in widest circulation was reported on a contact-centre industry page. That page described it as the longest time customers would wait on a call, and attributed it to a 2014 survey. The same page carries a later industry study of the same question, and the two findings are years apart and nothing like each other. Only the older one, about phones, travelled.
Measured where leaving is hardest
Look again at where these figures were collected. A hospital, where leaving means going home untreated. A tax helpline, where leaving means the question stays open. Those are queues people have powerful reasons to stay in, and they are the only kind anyone publishes numbers about.
Nobody has measured a queue that’s easy to walk away from. A café queue is about the easiest kind there is. No registration, no triage, and nothing lost by stepping out except a place in line and a coffee you can buy next door.
So read the walkaway rates in this article as floors rather than estimates. In a queue with nothing holding people in it, the share who leave is probably higher than any figure here. No published count says by how much.
The next question along, why a person leaves at all, has its own answers: why customers leave a queue.
Frequently Asked Questions
How long do customers wait before leaving?
Nobody has measured it for shops or restaurants. In hospitals, leaving climbs steadily with the wait rather than switching on at a point. In three emergency departments in Rhode Island, two more patients left per day for every 20-minute increase in the median stay of discharged patients (West JEM, Providence, 2025). Expect a slope, not a deadline.
What is queue abandonment?
It covers two different behaviours. Reneging is leaving a queue after joining it. A 2025 study defines it as a waiting customer leaving before receiving service (Int J Tourism Res, 2025). Balking is deciding not to join at all, usually after one look from the door.
What percentage of customers give up and leave?
There’s no figure for shops. In hospitals, published rates start at 0.15% of registered patients at a Tokyo children’s hospital, on 2014 to 2017 data (Cureus, 2024). They reach 12.12% near Naples, on 2019 to 2023 data (BMC Emerg Med, 2025). Those services aren’t counting the same behaviour: Northern Ireland counts attendances that left before treatment was complete (Department of Health, 2026).
Does telling people how long they will wait stop them leaving?
Not reliably, and the evidence is thinner than the advice suggests. The only recent test used made-up restaurant scenarios, not real queues, with 142 and 170 people in the United States. It found that the wording mattered: a coarsely worded wait drew more walkaways than the same wait put in finer units (Int J Tourism Res, 2025). Stated intentions, though, not what anyone did.
Are there any UK figures for people giving up and leaving?
There is one. During June 2026, 6.8% of Northern Ireland emergency attendances “left before their treatment was complete”, on a base that excludes two urgent care centres (Department of Health, 2026). Phone queues publish waiting times rather than a walkaway rate. HMRC reports what callers waited between April and June 2026, but not how many gave up (GOV.UK, 2026). For queues in shops, nothing.
Sources
- BMC Emergency Medicine, “Predicting patient risk of leaving without being seen using machine learning: a retrospective study in a single overcrowded emergency department”, published 15 July 2025, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC12261541/
- Cureus, “Factors Associated With Leaving-Without-Being-Seen in Pediatric Emergency Department Patients”, published 7 December 2024, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC11702983/
- Department of Health (Northern Ireland), “Emergency care waiting time statistics (April – June 2026)”, published 12 August 2026, retrieved 2026-08-28, https://www.health-ni.gov.uk/news/emergency-care-waiting-time-statistics-april-june-2026
- HM Revenue and Customs, “HMRC performance update: 2026 to 2027 quarter 1”, published 13 August 2026, retrieved 2026-08-28, https://www.gov.uk/government/publications/hmrc-performance-update-april-to-june-2026/hmrc-performance-update-2026-to-2027-quarter-1
- International Journal of Tourism Research, Baek, Ok and Lim, “How to Prevent Waiting Customers From Leaving: The Interaction Impact Between Granularity and Mindset on Reneging Behavior”, volume 27 issue 4, article e70102, published 21 August 2025, retrieved 2026-08-28, https://doi.org/10.1002/jtr.70102
- JAMA Network Open, “Trends and Hospital Factors in Emergency Department Patients Leaving Without Being Seen, 2015-2024”, published 21 May 2026, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC13195482/
- Medicine (Baltimore), “Characteristics of patients leaving the emergency department without being seen by a doctor: The first report from Türkiye”, published 15 November 2024, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC11575991/
- NHS England, “Supplementary ECDS Analysis Time Series, February 2023 Onwards” (workbook linked from this page), retrieved 2026-08-28, https://www.england.nhs.uk/statistics/statistical-work-areas/ae-waiting-times-and-activity/
- Ofcom, “Abandoned and silent calls”, retrieved 2026-08-28, https://www.ofcom.org.uk/phones-and-broadband/unwanted-calls-and-messages/abandoned-and-silent-calls
- Ofcom, “Telecoms companies up their game on complaint handling, but more to do to keep customers satisfied”, published 22 May 2025, retrieved 2026-08-28, https://www.ofcom.org.uk/phones-and-broadband/service-quality/telecoms-companies-up-their-game-on-complaint-handling-but-more-to-do-to-keep-customers-satisfied
- Postgraduate Medicine, Kadioglu and Özkan, “Operational and demographic predictors of leaving without being seen in a high-volume tertiary emergency department: a five-year case–control study”, doi 10.1080/00325481.2026.2696684, published June 2026, retrieved 2026-08-28, https://doi.org/10.1080/00325481.2026.2696684
- Production and Operations Management, Aksin, Gencer and Gunes, “How Observed Queue Length and Service Times Drive Reneging Behavior in Queues”, volume 34 issue 6, pages 1440 to 1457, published online 19 September 2024, retrieved 2026-08-28, https://doi.org/10.1177/10591478241286504
- Scientific Reports, “Investigation of emergency department abandonment rates using machine learning algorithms in a single centre study”, published 22 August 2024, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC11341825/
- Western Journal of Emergency Medicine, “Factors Associated with Patients Leaving Without Being Seen in a Canadian Emergency Department”, published 23 December 2025, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC12815564/
- Western Journal of Emergency Medicine, “Reduced Functional Bed Capacity Due to Inpatient Boarding Is Associated with Increased Rates of Left Without Being Seen in the Emergency Department”, published 26 November 2025, retrieved 2026-08-28, https://pmc.ncbi.nlm.nih.gov/articles/PMC12698166/