A branch visit, counter by counter

Branch traffic is not random. It arrives at lunchtime, on payday, and in the first three days of the month, and every branch manager knows it. The problem has never been prediction. It has been that a single line cannot express the difference between a customer depositing a cheque and a customer restructuring a loan.

The thirty-second customer behind the forty-minute customer

In an unsegmented queue, service time is a lottery. A customer with a two-minute transaction who happens to arrive behind three account openings will wait an hour, conclude the branch is badly run, and be right.

Segmenting at the door fixes this with no extra staff. The kiosk asks one question — what did you come to do — and issues a token from the matching pool. Quick transactions flow to tellers; long ones to advisors; priority customers to whichever is appropriate. The line does not get shorter. It gets sorted.

What segmentation actually needs

• Three to six categories, not twenty. A customer who cannot classify themselves in four seconds will pick wrong, and a wrong classification is worse than none.
• Plain names. “Cash and cheques”, not “Retail Transaction Services”. Name things by what the customer came to do.
• A transfer path. Some customers will still pick wrong. Moving a token between queues must take one click and must not send them to the back.

Let them stop standing

Once a token lives on the customer’s phone, the physical line stops being the queue — it is just furniture. Customers sit, use the branch’s own digital channels while they wait, or step next door for a coffee, and come back when the message arrives.

Two things follow. The branch feels calmer, which affects how staff work as much as how customers feel. And the floor space you were dedicating to a rope line becomes space you can use for something that earns.

Staffing to the curve, not to the roster

After a month of tokens, the peak stops being anecdotal. You can see that Tuesday between 12:15 and 13:40 needs a fourth teller, that the advisor desk is idle before 11, and that the first working day of the month is a different business from every other day.

That is a rostering input, and it is the single largest cost lever in a branch. Most branches are not understaffed. They are staffed evenly against demand that is not even.

What to measure

• Wait by segment. The number that matters to a customer is the wait for their service.
• Counter utilisation across the day. Idle time at 10 a.m. is the staffing you need at 1 p.m.
• Transfer rate. A high rate means your categories are named wrong.
• Score by desk. Branch-level satisfaction is an average that hides the one desk that needs support.