Why a Quiet Bakery Can Still Have a Queue
A good staffing plan accounts for the moments when several customers arrive together. An hour’s sales total can conceal those moments entirely.

In this story 6 sections
At the end of an hour, twelve transactions look manageable. At the counter it mattered that six came in after a bus let people off. One needed a cake boxed, two wanted the same item in different quantities, and one transaction took place while another customer was trying to ask where the line started. The hour’s average does not explain why there were five people waiting at one moment and nobody a few minutes later. To understand the queue, draw the hour instead of dividing it into twelve equal pieces.
A minute can contain several jobs
A customer may need help choosing, then a product needs to be fetched, then payment goes through. In a one-person shop the same worker also replenishes the counter. During a burst those jobs arrive in a different order than they would on a steady day. The person handing over a loaf cannot simultaneously ask the next person about a complicated order. A second worker standing beside the first helps only if something can usefully happen in parallel; two people sharing a tiny space and one terminal can get in each other’s way.
Draw one line for arrivals and a second line for finished transactions. The gap at a particular point is a count of people who have arrived but have not finished. A steep section in the arrivals line reveals a burst. A flat stretch in the completions line may mean that service has stalled or simply that nobody is present. Once the two lines are together, that ambiguity usually resolves. Divide the observation by time of day and purpose instead of trying to explain the business with a single average.
Choose what you want to improve
The average number of customers present, the rate at which they are served and their average time in the system are related under the conditions of Little’s Law. The averages are useful when the same system and observation boundaries are used throughout. They do not promise that any individual customer will be served in the average time. One policy might shorten a typical wait and lengthen an occasional very bad wait. Another may take the most uncertain order out of the main line, making the overall pattern calmer without a dramatic jump in completed sales.
Examples of adjustments include moving packaging away from the payment station, confirming prepaid pickups at a separate point, or timing cleanup and preparation around known surges. There is a cost to each change. One line may be easier to understand than two. A worker moved to help at the front may leave a different task unfinished. Observe why people stop moving before proposing a separate lane or an extra till.
Check tomorrow as well as today
For a concrete comparison, follow one fifteen-minute window on two different days. Note when each person joins a waiting group, when staff first take their order and when they leave with it. Those are three times. If a person orders immediately and waits for a drink to be made, the ordering queue was short and the total visit took longer. Another person might wait to speak but receive an already prepared item quickly. The manager of a busy room could choose to improve one of those intervals first. Recording just the time printed on a receipt might conceal the part the customer noticed most.
Draw the two waits beside one another for a handful of people, using the same starting point in each account. A person who had the longest total visit may not have queued the longest before speaking. A policy that moves preparation elsewhere can make the entrance quieter without improving pickup times. A policy that prepares common items ahead could reduce the time between ordering and handover. The two experiments give different predictions. They should be compared using the part of the visit each one actually changes.
A short observation explains a particular shift. It can be a poor forecast for a holiday or an unusual weather day. Make a note about what was different before comparing results. Report a range of waits, especially the longest periods in which customers arrived faster than they left. If the change is about telling people where to stand, success might look like fewer blocked doors rather than a higher rate of transactions. A useful schedule starts with the shape of the day, including the minutes that its average smoothed away.
The longest waits can disappear from the receipts
A person can come to the door, see a crowd and leave before taking a place in line. Another can leave after standing for a while. Neither makes a sale. A report based on completed transactions will miss both, and can make some measures of time appear better during a difficult period. There is a similar problem at the end of an observation. People still waiting at that moment have not finished their waits. Give them a separate mark and keep the comparison period clear rather than allowing an incomplete count to stand for an ordinary result.
Counting departures without a purchase is harder than counting receipts because a shop often has several reasons for being entered. Someone might be looking for another person or checking the menu. You can still learn from readily observable events. “Three people left after waiting at the counter” says more than assigning motives to everybody who went past the door. A manager could then investigate whether the group had a clear sense of who would be served next. Improving the signs may make the room easier to use even when the average time spent making each order stays the same.
A well designed measurement does not need to capture a personal detail about a customer. It needs enough consistency to show the differences that matter. Write small anonymous counts beside times and leave names, orders and payment information out of the notes. Record interference from an unusual event, such as a machine being out of service. Be explicit if the data come from a single visit. An honest brief observation can be more helpful than an exact figure without a description of the conditions under which it was taken.
A second station is useful only when it can do a separate job
Suppose a bakery assigns somebody to prepare boxes and fetch products during the usual rush. If the same worker also has to use the only payment terminal at exactly the same moments, the proposal may simply move where people stand. Write a short order from memory, then trace who can carry out each of its tasks while someone else handles the next person. Packing might need its own surface and a place for an order awaiting handover. A separate position can increase the rate at which work is finished when it removes competition for a shared tool or stretch of floor. Adding a position that has to compete for the same resource can produce a much smaller change.
There are cases where an extra position is valuable for a short window. Moving unhurried preparations into a quieter time is one option. Keeping a person free to respond during a predictable burst is another. Both choices can cost something elsewhere. Make a separate note of duties displaced by the proposed change, including the next rush that will arrive if preparation falls behind. A shop can improve one fifteen-minute interval and make the rest of a shift worse. The better comparison accounts for the service and all of the work that will recur after the queue passes.
After the trial, ask whether customers now reach the service point in a sensible order and whether work waiting to be completed has a clear place. Follow one busy period rather than relying solely on an average. Spare capacity is easiest to understand in the seconds when something unplanned arrives: a complicated order, a payment retry or several people entering together. If the arrangement absorbs those moments without blocking exits or putting all the work on one station, there is a practical improvement that a single average would struggle to describe.
Read a short return period with care
Imagine the first ten minutes after a group arrives produce twelve orders, while the following ten minutes produce four. If the counter completes eight in each interval, it handles exactly the sixteen orders over twenty minutes. There will still be a group of four waiting at the boundary between the intervals, assuming the counter began free and all orders take one place in the same series. Looking only at totals makes the work appear balanced. The group experienced a period in which arrivals got ahead of the ability to finish them and then a period in which the shop caught up.
The example explains why smoothing a predictable burst can accomplish something even when capacity does not change. A prepaid pickup option can shift some of the work if the preparation really happens earlier. Clearer signs can spread a confused group into a line that reaches the correct place. One change modifies when orders arrive for handling. The other modifies how much space waiting consumes and how easily people know what to do. Give each proposal an outcome that can be observed. The work of reducing waiting and the work of making a wait easier are both real, and a receipt log may show only one of them.


