Every café owner knows the morning rush as a good problem. It is the busiest ninety minutes of the day and the till reflects it. What the till does not show you is the other number: the people who looked through the window, saw the queue, and kept walking.
Those customers are invisible by definition. They leave no transaction, no receipt and no complaint. They are also, in most venues, the largest single source of lost revenue in the day.
Two words worth knowing
Queue theory has names for what your customers are doing, and they are useful because they are different behaviours with different fixes.
- Balking — arriving, seeing the queue, and not joining it. The customer never enters. You never know.
- Reneging — joining the queue and then leaving before being served. You see this one, usually as someone quietly stepping out of line.
Both spike at exactly the time your product is most wanted. Both are worst with the customers you most want, because a commuter with a train to catch has the least tolerance for a wait and the highest visit frequency.
The arithmetic
Run it on your own numbers. Take an average ticket — call it £4.50 for illustration — and estimate how many people balk during the peak. Most owners guess low; stand outside for two mornings and count, and the number is usually a surprise.
| Walkaways per morning | Per week (6 days) | Per year, at £4.50 |
|---|---|---|
| 3 | 18 | ~£4,200 |
| 6 | 36 | ~£8,400 |
| 10 | 60 | ~£14,000 |
These are illustrative figures, not a claim about your venue — the point is the shape. Three people a morning does not feel like a problem. Over a year it is a member of staff.
And the loss compounds, because a balking customer is rarely a stranger. They are a regular in the making who has now learned that your venue is unreliable when they are in a hurry, and they only need somewhere else to work once.
Why more staff does not fix it
The instinct is to put another person on. It helps, but less than it should, because the bottleneck usually is not the machine — it is the ordering step.
Taking an order is conversational and slow: what would you like, what size, which milk, anything else, and a payment. During a peak, a second person on the till often just moves the queue to the same espresso machine. You have paid a wage to reshape the queue rather than shorten it.
The fix is to move the ordering step out of the peak, not to process it faster. An order taken at 07:52 by someone on a train is an order that costs your counter nothing at 08:05.
What order ahead changes
- Balking stops mattering for anyone who has your link. They ordered before they saw the queue. The queue is now somebody else's problem.
- Demand arrives with a few minutes of warning. You get preparation time you previously did not have.
- The peak flattens slightly. Some people order at 07:40 for 08:10 simply because it is easier.
- Staff stop being interrupted. Fewer conversations at the till, more drinks made.
- The order is more accurate. The customer typed it. Nobody misheard "oat" over a grinder.
The collision you have to plan for
Order ahead creates one new operational problem and it is worth naming before you switch it on: two queues competing for the same barista.
If you make every order-ahead ticket the moment it lands, the counter queue slows down and the people standing in front of you get annoyed. If you ignore the screen until the queue clears, the order-ahead promise breaks. Neither is acceptable.
- Work to the collection time, not the arrival time. A ticket for 08:15 does not need starting at 08:04.
- Batch what batches. Two flat whites for different tickets are still two flat whites.
- Give collection a dedicated spot so it does not consume counter time. Handing over is seconds; finding the drink is not.
- Widen the preparation window during the peak. An honest fifteen minutes is better than a broken five.
- Cap it if you must. It is entirely reasonable to switch order ahead off for an hour when the venue is beyond capacity.
Measuring whether it worked
Total revenue is too noisy to tell you anything month to month. Two better signals:
- Peak-hour transaction count. If the same ninety minutes now serves more people, the queue was the constraint and you have removed part of it.
- Repeat rate among order-ahead customers. The habit is the return. Someone who has ordered ahead three times has restructured their morning around you.
The second one is the one that pays. Removing a queue is worth something once; becoming the default stop on somebody's walk to work is worth something every weekday for years.
Coffeeloc gives your regulars an order-ahead page for pickup, and a stamp card that gives them a reason to make it a habit.