Wait Times: What's Normal, What's Not, and How to Measure Yours

Ask the front desk what the average wait is and you will hear a number delivered with total confidence: "usually about twenty minutes." Ask where that number came from and the conversation ends. It was not measured. It is a feeling, shaped by whichever recent shift was most memorable, and it is the number the business repeats to customers, puts in review replies, and staffs against.

This piece is a benchmarking guide with an honest problem at its centre: outside healthcare, almost nobody publishes wait-time data, so the benchmark that matters most is the one you build yourself. What follows is the public record such as it is, the reason most operators cannot state their own average, the difference between the wait on the stopwatch and the wait in the customer's head, and a measurement path that starts with a notepad.

The only published benchmarks worth citing

Walk-in healthcare is the one sector with real published data, and it comes from Medimap's Wait Time Index. The average Canadian walk-in clinic wait was 68 minutes in 2023. Ontario averaged 59 minutes, up from 25 minutes just a year earlier, and Toronto came in at 72 minutes, the longest in the province. Those three figures are close to the entire public record. If you run a salon, a pharmacy counter or a repair shop, there is no Medimap for you, and any precise "industry benchmark" a vendor quotes without a named source was probably invented to sell you something.

One more published number is worth holding onto, from the Urgent Care Association: a left-without-being-seen rate under 2 percent is treated as the mark of a well-run centre. That number travels beyond healthcare because it reframes the question. The interesting benchmark is not how long people wait. It is how many stop waiting.

So what is normal? For a Canadian walk-in clinic, roughly an hour, which should tell you how little comfort there is in being normal. For everyone else, normal is unknowable from the outside. Your own trailing average, measured over a few weeks, is the only baseline you will ever get. The good news is that it is cheap to build.

Why you probably can't state your own average

The paper sign-in sheet is a list of names, not a record of time. It does not know when anyone arrived, when they were called, or how long the person at the counter took. It certainly does not record the person who read the room, did the math on the crowd, and left without writing anything down. Paper records nothing you can average.

Memory fills the gap, and memory is a terrible instrument. You remember the Saturday the line went out the door; you do not remember forty ordinary Tuesdays. Guessed service times run short too, because the answer a barber gives for how long a cut takes never seems to include the cleanup, the payment, or the chat. So the confident "about twenty minutes" is a story. It might even be a true story. You have no way to know.

Actual wait and perceived wait are different numbers

A stopwatch measures one wait; your customer experiences another. David Maister's old point about uncertainty explains most of the gap: a wait you can watch move is a different thing from a wait you can only endure, whatever the clock says. Twenty-five minutes in the car with a live queue position on a phone and twenty-five minutes in a hard chair staring at a door are the same number and completely different experiences. Only one of them shows up in your reviews.

This cuts both ways for measurement. The thing that converts an uncertain wait into a known one is an estimate, and an estimate built on a guessed service time is a lie with a progress bar. Serious efforts to reduce wait times work both dials at once, shrinking the real wait and the felt one, but the measured number has to come first.

The measurement path: three timestamps

Every wait metric you will ever want derives from three timestamps per customer: when they joined the line, when they were called, and when you finished with them. Join to called is the wait time. Called to finished is the service time. That is the entire data model.

The manual version costs a notepad. On a normal day, log those three times for twenty consecutive customers. Average join-to-called and you have your first real wait figure. Average called-to-finished and you have the service time that should sit behind every estimate you give anyone. Repeat on a different weekday, because Saturday and Tuesday are different businesses. Two hard limits: a notepad still cannot see the customer who took one look at the crowd and left, and two days is a sample, not a distribution.

Once you have a measured service time, you can project waits forward instead of guessing. The standard formula is people ahead multiplied by average service time, divided by providers working, and you can run it against your own numbers with the free wait-time calculator before you change anything about how you operate.

The four numbers worth reading every morning

Measurement only matters if it changes a decision, so keep the daily view small. Four numbers cover the decisions a walk-in business actually makes:

• · Average wait time (join to called), the customer-facing number and the one to put live estimates behind
• · Average service time (called to finished), the capacity number; when it drifts, every estimate built on the old figure quietly goes wrong
• · Show rate, the share of people still there when their turn came; its inverse is the walkaways the wait cost you
• · Busiest hour, the staffing answer, because you cannot schedule to a rush you have never located

A digital queue produces all four as a by-product, because it timestamps every join, call and completion without anyone holding a stopwatch. LineMarshal's analytics tab reports them over the last 7, 14 or 30 days on every plan including the free one, alongside customers served, no-shows and total people, and it nudges you when your measured service time drifts from the number you configured, which is exactly the drift that poisons estimates.

Frequently Asked Questions