How to Quote a Wait Time
Take an invented phone repair counter in a shopping centre, on a Saturday afternoon. A customer puts down a phone with a cracked screen and asks how long it will take. The person behind the counter looks at the phones already waiting and gives the easy answer: “Shouldn’t be long.”
That’s a quote, even if it doesn’t sound like one. To the customer it might mean time for a coffee. To the technician it might mean the rest of the afternoon. Neither of them has said which.
A better quote is short and specific. Give the time that most of your recent customers were served within, not the average. Say it in plain minutes, and say when you’ll next update them. If the quote is going to be missed, tell the customer before it runs out, not after.
The sections below take each part in turn. They cover which figure to say, whether padding helps, how to word it, what to show when there’s no honest number, and when to go back.
Key Takeaways
- Quote the time most of your recent customers were served within, not the average. An average can sound fine while many people wait longer.
- Running late against a quote cost something in every study here that measured it. Finishing early is no reliable win, and in one experiment it left people less satisfied with the quote’s accuracy.
- Say the wait in plain minutes. In recent scenario experiments, coarser units for the same wait made people likelier to say they’d leave, and less willing to join.
- If you can’t give minutes, show the queue or the progress instead. If the quote will be missed, tell the customer before it runs out.
Which number should you say out loud?
Say the time that most of your recent customers were served within, with the typical wait beside it if you can. Don’t say the average. An average can sound reasonable while a large share of people wait longer than it.
HMRC’s phone lines show how large that share can be. In September 2026, the average wait to reach an HMRC adviser was 8 minutes 30 seconds, and 36.3% of calls that advisers handled had waited more than 10 minutes.
Two details sit behind those figures. HMRC’s average starts the clock only once a caller has finished with the automated menu, so time spent picking options isn’t in it. And the 10-minute share covers only calls that an adviser went on to answer.
That average sounds manageable. Quote it to every caller, though, and on more than a third of the calls that reached an adviser, the wait would have run past it. A figure that nearly every one of those calls came in under would have to sit beyond 10 minutes, further out than the average.
Hong Kong’s public hospitals have shown waits both ways on one web page. Until October 2025, the online A&E page showed a single estimated waiting time. Researchers who studied that system described the posted figure as the 95th percentile of recent waits over a three-hour window. Only the longest recent waits went past it. They argued, from a model, that a less extreme and more recent figure would serve patients better.
The page now breaks waits down by triage category. For the three lower-priority ones (urgent, semi-urgent and non-urgent), it gives two figures instead of one. One is the time half of waiting patients are seen within, and the other is the time most are seen within. For the two least urgent, the estimate is rounded up to the nearest half hour. Rounding up, rather than to the nearest, means the posted time errs on the long side.
A counter can do the same in one breath. “Usually about twenty minutes, could be forty” gives the typical wait and the long end together, and the numbers in it are only an illustration. It’s also a way of giving a range rather than a single time.
Both numbers need to come from real waits, not memory. Working out those two figures from your own records mostly means writing down when each wait started and when it ended.
Should you pad a wait time quote?
A small cushion that keeps most customers on the early side, yes. A large pad, no. In every study below that measured lateness, running later than the quoted time cost something. Finishing early gave mixed results, and in one study it cost something too.
Four studies looked at what follows when a promised time is beaten or missed. None of them measured a queue at a counter: all four are about deliveries or online waits, and the setting column says which.
| Study | Setting | When it ran later than quoted | When it ran earlier than quoted |
|---|---|---|---|
| Pavlov, Mogre and Olsen, Journal of Retailing and Consumer Services | Online waiting experiment, United States | Overall satisfaction fell | Overall satisfaction neither reliably rose nor reliably fell |
| Mao, Ming, Rong, Tang and Zheng, Journal of Operations Management | Meal delivery platform, China, transaction records | Customers placed fewer future orders | Only limited benefits |
| Xie, Sharma, Mehra and Aziz, Information Systems Research | Food delivery marketplace, transaction records | The study’s focus was showing a cautious estimate | Beating a cautious estimate brought more repeat orders |
| Nozari and Alavi, Production and Operations Management | Parcel deliveries from a major online shopping platform | More customers left a rating, and the ratings were lower | More customers left a rating, and ratings were lower in every product category except food and beverages |
Early is where they disagree. Xie’s team found that beating a cautious estimate brought customers back, both to the platform and to the same restaurant. Pavlov’s and Mao’s teams found it gained little or nothing. Nozari and Alavi found early parcels drew lower ratings in every category except food and beverages.
Pavlov’s experiment adds a detail that matters at a counter. People were less satisfied with the quote’s accuracy when the wait missed it in either direction. So in that experiment, finishing early earned no reliable credit. It counted against the quote, just in the friendlier direction.
A big pad has a cost at the front end too. Xie’s team found the cautious estimate lost some orders up front, though repeat orders more than made up for it on that platform.
Where the cushion comes from matters as much as its size. Build the quote from how long people actually waited, not from what you told them last time. A padded quote that becomes tomorrow’s starting point carries its padding forward, and the next cushion lands on top of it.
Minutes, not quarters of an hour
Give the wait in plain minutes. The same wait can be put in finer or coarser units. In two recent sets of scenario experiments, the coarser version made people less willing to stay and less willing to join.
In the first, people were given the same wait either in minutes (15 to 30) or in fractions of an hour (a quarter to half an hour). The coarser wording led more of them to say they would leave the queue before being served, which researchers call reneging.
The second, on a restaurant waitlist, looked at whether people join at all. Declining to join is called balking. Finer-grained first estimates raised people’s expectations of being served sooner than stated, and fewer of them balked. Even if later updates change expectations, the authors argue, the first estimate still matters, because it has already shaped who joined the queue.
Both are scenario experiments. They record what people said they would do, not what anyone did at a real counter.
In practice, if the honest figure is twenty-five minutes, to take an illustrative case, say twenty-five minutes rather than “half an hour or so”. The step should suit the size of the wait. Hong Kong’s half-hour rounding applies to waits that are stated in hours. At a counter where the whole wait is under an hour, minutes are the matching step.
What if you can’t put a number on it?
You can still tell people something true. Three public bodies show how, without promising an exact time. They show the queue, show the progress, or give a ceiling and say how the person will hear.
| What they show instead of an exact time | Who does it | How it might look at a counter (illustrative) |
|---|---|---|
| The queue: patients assessed and waiting, the usual number arriving, and the beds or treatment spaces | NSW Health, Australia, through an online search tool | “Three repairs ahead of yours” |
| Progress: the date of the oldest application now being worked on, updated weekly | Legal Aid Agency, on GOV.UK | “Now working on the 10:40 drop-offs” |
| A ceiling and how the person will hear: “You’ll get an email once your results are ready. This can take up to 15 minutes.” | Scottish Government Design System, in a pattern for web forms | “Up to an hour; you’ll get a call” |
Each one fits a different kind of wait. Showing the queue suits a line where the order can change. The Service NSW page is plain about the limits. “The most seriously ill patients are seen immediately”, and the guide “doesn’t indicate a specific timeframe or order in which a patient will be seen”. A count of the people ahead stays true when a promised time wouldn’t.
Showing progress suits work taken in order of arrival. The Legal Aid Agency publishes how far through the pile it has got.
A ceiling suits a one-off job of unknown length, as long as the next contact comes with it. The Scottish pattern is written for web forms, not counters. Its rule still travels: when the result isn’t known yet, tell the person “what happens next and when”.
In the repair shop from the opening, the answer might mix two of these: how many repairs are ahead, and a promise to ring when this one is done. That’s more use to the customer than “shouldn’t be long”, and neither part depends on a guess.
Re-quote before the time runs out
Go back to the customer before the quoted time passes, not after it. A new time given early is just a revised quote. The same news given late, after the customer has been watching the clock, lands as a broken promise.
UK Visas and Immigration puts the rule in writing. Its guidance says: “You will be contacted within the current processing times if your application will take longer”.
Hong Kong’s A&E page does the same job on a screen. According to The Standard, the page shows a “light bulb” indicator when a department is handling large numbers of high-priority patients. It warns that lower-priority waits may be longer while that lasts.
Precision raises the stakes. A 2026 meta-analysis found that the gap between what customers expected and what they got matters more when they had clearer information. An exact quote is about as clear as information gets. The sharper the quote, the harder a miss lands, which is the starting point for what a missed quote does to reviews.
Keeping people informed also matters for its own sake. At emergency departments in Newfoundland and Labrador, Canada, patients who said they were kept informed about delays reported being better satisfied. The authors link being kept informed with patients feeling “less forgotten”. The survey ran from 2021 to 2023, and it shows a link, not proof of cause.
When to go back to someone who is waiting:
- Before the quoted time, if you’re going to miss it. Give the new time while the old one still has some life left in it.
- When something ahead of them changes. An urgent job arrives, or someone goes home ill.
- At the time you said you’d update, even with no news. A check-in with no change still shows the customer they haven’t been forgotten.
- When the wait gets shorter. People who planned around the old time may wander off and miss their turn.
The part of the quote you control
Nobody at a counter controls how long the job ahead of the customer takes. A part arrives late, or a simple fix turns out not to be simple. A quoted time is a forecast, and forecasts miss.
The promise to come back is a different kind of thing. Something like “I’ll let you know by half past” depends only on the person who said it. It can be kept however the repair goes, which makes it the one part of the quote that never has to be wrong.
That’s where the effort belongs when something has to give. Get the number as close as your records allow, and accept that it will sometimes slip. Then treat the time you said you’d come back as fixed. A business that misses forecasts but always turns up when it said it would is giving customers something they can plan around.
Frequently Asked Questions
How much should you add to a wait time quote?
No fixed amount works everywhere, because the right cushion depends on how much your waits vary. The two figures from your own records show that. If the long end sits close to the typical wait, a small cushion covers most customers. If it sits far beyond, the cushion has to be bigger, or the quote should be a range.
How many past waits do you need before you trust your own figure?
More than a handful. NHS Wales greys out any hospital specialty with fewer than 200 patient pathways, because there isn’t enough data to calculate reliable waiting times. That threshold is for hospital waiting lists, not a target for a counter. The principle carries over, though. Don’t build a quote from a few memorable customers, and compare like hours with like, so a slow Monday isn’t setting Saturday’s quote.
Do customers want the likely wait or the worst case?
Early evidence points to the long end. In a 2026 preprint that has not yet been peer-reviewed, people deciding what to learn about a wait overwhelmingly sought information about the long end. Its authors suggest that when the aim is to inform, providers should share the long-end figures first. Treat it as a pointer until it has been reviewed.
Sources
- HM Revenue and Customs, HMRC performance data 2026 to 2027: quarter 2, published 12 November 2026
- HM Revenue and Customs, definitions for metrics used in the performance report, published 12 November 2026
- The Standard (Hong Kong), Hospital Authority revamps A&E waiting time system to show category-based estimates, published 15 October 2025
- Manufacturing & Service Operations Management, Park, Ouyang, Wang, Savin, Leung and Rainer, Patient Sensitivity to Emergency Department Waiting Time Announcements
- Hospital Authority (Hong Kong), Data Specification for Accident and Emergency Waiting Time
- Journal of Retailing and Consumer Services, Pavlov, Mogre and Olsen, Ex-post customer reference effects of lead-time quotation in an online waiting experiment, volume 94
- Journal of Operations Management, Mao, Ming, Rong, Tang and Zheng, Faster Deliveries and Smarter Order Assignments for an On-Demand Meal Delivery Platform, published 20 February 2025
- Information Systems Research, Xie, Sharma, Mehra and Aziz, Strategic Expectation Setting of Delivery Time on Marketplaces, volume 35 issue 4
- Production and Operations Management, Nozari and Alavi, Please Arrive On Time: The Impact of Early and Delayed Delivery on Product Ratings, published 3 April 2026
- Journal of Hospitality & Tourism Research, restaurant waitlist experiments on how a first wait estimate is stated, published 4 September 2026
- International Journal of Tourism Research, two experiments on stating a wait in finer or coarser units, published 21 August 2025
- Service NSW, Check wait times for major NSW hospital emergency departments, last published 7 May 2026
- Legal Aid Agency, Civil processing dates, GOV.UK, last updated 17 November 2026
- Scottish Government Design System, Form submission results page, last updated 1 February 2024
- UK Visas and Immigration, Visa processing times: applications inside the UK, GOV.UK, last updated 9 September 2026
- Journal of Service Research, Lin and Agarwal, Dual Perspectives of Customer Wait Time Perception: A Conceptual Framework and Meta-Analysis, published 24 June 2026
- Journal of Patient Experience, Manukumar and colleagues, They Just Don’t Want to Feel Forgotten, published April 2025
- NHS Wales, 111 Wales Planned Care, Frequently Asked Questions
- arXiv, Kagan, Hyndman and Davis, Chasing Tails: How Do People Respond to Wait Time Distributions?, preprint submitted 5 February 2026