How Many Tellers Should I Schedule Each Day at My Bank Branch in 2026?

What branch staffing actually measures and why habit fails

The question "how many tellers should I schedule each day" sounds like a headcount question. It isn't. It's a capacity question wearing a headcount costume, and the two behave very differently once you look at how a branch lobby actually fills up.

Capacity in a branch is the product of three things: how many service windows are open, how fast each window clears a transaction, and how customers arrive across the day. Arrivals are not smooth. They come in bursts — a spike at open when people who waited overnight come in, a swell through the lunch window when working customers use their break, and a heavy Friday tied to pay cycles and the settlement of the week's business. Layer in the first and fifteenth of the month, when benefit deposits, rent checks, and payroll runs land together, and you get days where volume is 30 to 50 percent above the branch's own average.

Scheduling by habit — "we've always run six" — ignores every one of those variables. It sets a flat line across a curved demand pattern. The result is the failure mode every regional manager recognizes: idle tellers standing at empty windows on a dead Tuesday afternoon, and a line snaking to the door on Friday at 12:15 with the same six people on the schedule. You are simultaneously overpaying and underserving, in the same week, in the same branch.

The alternative is to anchor the count to something the branch actually produces. In retail, that anchor is gross profit per employee per day. Banks flinch at the phrase because a branch doesn't ring a register, but the concept translates cleanly. A branch's daily contribution is the net interest margin it earns on the deposits and loans booked or held there, plus fee and service income, minus the cost of funds — allocated by day of week. That number exists in your finance reporting. It may be called branch contribution, net revenue, or direct margin. Whatever the label, it is the money the branch made that day, and it is the honest denominator for a staffing decision.

Once you have it, the method is a division problem. Leadership agrees on the gross profit an average teller or banker should produce on an average day serving an average number of customers. Say $400. That is a floor, not a ceiling — the person who wants to grow hits $400 doing ordinary work and then digs for the next $400 with a real referral. Divide the branch's average daily contribution by that target and you have your count. A $2,000 Wednesday needs five. A $3,600 Friday needs nine.

Why this matters more than it used to: branch transaction volume has been declining for years as mobile deposit, ACH, and card payments absorb routine work. The branches that survived that shift didn't survive by cutting tellers uniformly. They survived by re-cutting the schedule so the remaining hours landed where value still concentrates — advisory conversations, account openings, small business deposits, loan applications, and the fraud and dispute work that customers refuse to do on a phone. The staffing math is what tells you which hours those are. A flat schedule buries the signal; a gross-profit-derived schedule surfaces it.

There's a second-order effect worth naming. When the count comes off a shared number instead of a manager's preference, the political fights stop. Nobody argues that their friend deserves the easy Tuesday shift, because Tuesday's count came from Tuesday's contribution. The yardstick is the same for leadership, the branch manager, and every person on the line. That transparency is often worth more than the labor savings.

Building the schedule step by step

Here is the sequence, start to finish. It takes a focused afternoon the first time and about an hour a month thereafter.

Step one — set the per-person daily target. Get your leadership team in a room and agree on one number: the gross profit an average teller or banker should produce on an average day. For most branch environments this lands well above a retail floor, because every transaction and referral carries real margin. Say the number out loud to the team so it isn't a secret spreadsheet input: "If you show up, take care of an average number of customers, and give average service, you should produce no less than $400 a day." Set it too high and you'll chronically understaff and burn people out. Set it too low and you'll pad the line. Sanity-check it against your fully loaded hourly cost — if a teller costs roughly $22 to $30 per hour loaded, an eight-hour shift costs $176 to $240, so a $400 target implies a healthy contribution multiple rather than a break-even one.

Step two — pull gross profit by branch by day of week. Take a trailing three to six months and average each branch's daily contribution by weekday. Three months is the minimum for stability; six smooths out seasonal noise. You want a table that reads: Midtown Monday $2,800, Tuesday $1,800, Wednesday $2,000, Thursday $2,200, Friday $3,600, Saturday $1,400. Do not average the whole week into one number — that's the exact information loss that produces flat schedules.

Step three — divide. Contribution divided by target equals staff count for that day. Midtown Wednesday: $2,000 ÷ $400 = 5. Friday: $3,600 ÷ $400 = 9. Round to whole people and apply a floor — most branches cannot legally or operationally open with fewer than two staff on the floor for dual control and cash security, so a slow Saturday that math says needs 3.5 people gets 4, and a branch that computes to 1.8 still gets 2.

Step four — place the shifts against the hourly demand curve. The division tells you *how many*. Hourly transaction data tells you *when*. Pull transaction and referral counts by hour for each branch over the same trailing window. Look for the shape: the open rush, the lunch swell, the late-afternoon fade, the Friday amplification. Then build shifts to match — full coverage at open, double coverage through the lunch window, your heaviest crew Friday, and a deliberately light 2 p.m. Tuesday instead of a full bench standing around.

Step five — layer in breaks, floats, and cross-training. A nine-person Friday is not nine people all day. Breaks and lunches remove roughly one person-hour per employee per shift, so nine scheduled bodies deliver about eight effective windows. Stagger breaks so you never drop below your lunch-window minimum. Keep one or two cross-trained platform bankers who can open a window when the line hits a threshold — many branches use a simple rule like "three people waiting for more than two minutes, a banker opens a window."

Step six — review monthly and re-cut. Contribution shifts with seasonality, local employers' pay cycles, branch consolidations, and construction detours. Re-pull the by-day table every month and adjust. This is where most implementations die: people run the math once, love the results, and never refresh it. Six months later the schedule is habit again, just a different habit.

A note on the queueing side, because it's the formal complement to the gross-profit method. Erlang C is the standard model for figuring out how many servers you need to hold a wait target given an arrival rate and an average service time. If your branch sees 30 arrivals per hour and average transaction time is four minutes, you need roughly two person-hours of service per hour of clock time — but to hold a five-minute wait target with random arrivals, you need meaningfully more than that, because randomness punishes tight capacity. The practical takeaway from queueing theory is counterintuitive and important: as utilization climbs past about 80 percent, wait times don't rise linearly, they rise sharply. A branch running its tellers at 90 percent busy will have wait times several times longer than one running at 70 percent, for a headcount difference of one person. Use the gross-profit division for the count and the queueing intuition as the sanity check on peak hours.

Costs, ranges, and what the numbers typically look like

Concrete ranges, so you have something to compare against.

Headcount. A small community branch handling 100 to 200 transactions a day typically runs two to four tellers, with a manager and one or two platform bankers. A mid-size branch at 300 to 500 daily transactions runs three to six tellers on the line, flexing to seven or eight on peak days. A high-volume urban or commercial branch handling 700-plus transactions, especially with business deposit customers, can justify eight to twelve. Those are the outer bands; your own contribution math should place you inside them, and if it places you far outside, that's a signal to check your inputs rather than to distrust the method.

Labor cost. U.S. bank teller wages generally sit in the mid-teens to low-twenties per hour depending on market, with fully loaded cost — benefits, payroll tax, training, and turnover amortization — typically running 1.25 to 1.4 times base. Turnover in teller roles is historically high, which means every schedule that chronically overworks people carries a hidden replacement cost. Recruiting and training a replacement teller commonly runs several thousand dollars once you count the vacancy, the hiring time, and the ramp.

The cost of guessing. Run the arithmetic on a single overstaffed weekday. Two extra tellers on a Tuesday, eight hours each, at $25 loaded, is $400 a day. Fifty-two Tuesdays is $20,800 a year, at one branch, for one day of the week. Do it on Tuesday and Wednesday across a twelve-branch footprint and you are into six figures of pure waste. The mirror-image cost is harder to see but larger: understaffed peaks produce abandoned visits, and an abandoned visit at a branch is frequently an account-opening or referral conversation that never happened.

Wait-time targets. The common operating target is under five minutes average wait during peak, with a tail target — something like 90 percent of customers served within eight to ten minutes. Beyond about eight minutes, abandonment climbs and satisfaction scores fall off steeply. Track both the average and the tail; the average hides the bad days.

Utilization targets. Healthy teller utilization during the busiest two-hour window is roughly 60 to 75 percent. Under 50 percent means you're paying for idle capacity. Over 85 percent means the queue is doing your capacity planning for you and your people are burning out. Measure it during the peak window, not across the whole day, or you'll flatten the very signal you're trying to read.

Tooling costs. Off-the-shelf workforce scheduling runs from free to enterprise. Homebase (joinhomebase.com, the U.S. employee-scheduling platform — not the unrelated UK home-improvement retailer of the same name) prices per location rather than per employee, which favors institutions running several small branches with large part-time rosters. When I Work and Deputy price per user per month in the low-single-digit-dollar range for base scheduling, more for time-and-attendance and compliance modules. Enterprise workforce management from UKG is quoted rather than listed, carries a multi-month implementation, and is the right answer only when you have hundreds of branches, union rules, or multi-state predictive-scheduling compliance to satisfy. None of these tools produce the count for you. They execute a count you bring. Confirm current pricing on each vendor's own site before budgeting.

Timeline. Expect two to four weeks of data collection if you don't already have hourly transaction reporting, one afternoon to build the first contribution-by-weekday table, one to two weeks to publish and stabilize the new schedule, and a full quarter before the labor and wait-time numbers settle enough to judge the change. Do not judge it in week two — the first week of any schedule change is noisy with swaps and complaints.

Where branch teams get this wrong

Averaging the week. The single most common error. A branch that does $14,000 a week and divides by five days gets $2,800 a day and schedules seven people every day. Monday and Friday are underwater; Tuesday and Wednesday are padded. The weekly average is the enemy of the daily schedule.

Confusing transaction count with value. Two hundred check cashings and two hundred new-account conversations are not the same day, even though the transaction counter reads the same. Value-weight your view. This is exactly why the gross-profit denominator beats a raw transaction denominator — a branch whose volume is falling but whose contribution is holding is a branch that shifted toward higher-value work, and cutting its staff on volume alone would be a mistake.

Ignoring the tail of the distribution. Managers optimize for the average day and get ambushed by the first-of-month, the day the local plant runs payroll, the day after a holiday closure, the Friday before a long weekend. Build a small named list of known spike days and pre-staff them. It costs a handful of extra shifts a year and prevents the worst customer experiences you'll have.

Treating the schedule as a fairness document. Once a schedule becomes a mechanism for rewarding favored staff with easy shifts, the math is dead. The defense is transparency: publish the contribution-by-day table and the per-person target so the count is visibly derived rather than decided.

Scheduling only tellers. The teller line is one component. Platform bankers, the branch manager, and any specialists share the lobby's load. A schedule that staffs the line correctly and leaves the platform empty during the lunch window just moves the queue. Model the branch, not the window row.

No overflow trigger. Even a good schedule meets a bad hour. Without a written rule — "at three waiting or two minutes elapsed, a banker opens a window" — you rely on someone noticing and choosing to act. Written trigger, everyone knows it, no judgment call required in the moment.

Skipping break math. Nine scheduled people is not nine windows. Breaks, lunches, cash drawer counts, and end-of-day balancing consume real hours. Count effective coverage, then stagger breaks against the demand curve so nobody's lunch lands at 12:15 on a Friday.

Never re-cutting. The method is a habit-killer only if you keep running it. Run it once and it becomes a new habit within two quarters. Put the monthly re-pull on a calendar with an owner's name attached.

Cutting to the bone on service and calling it efficiency. There is a floor below which you're not lean, you're just bad. If wait times routinely exceed eight to ten minutes, if utilization sits above 85 percent, if abandonment is climbing, or if teller turnover is rising — the schedule is too thin regardless of what the labor line says. The savings show up in this quarter's P&L; the attrition of both customers and staff shows up in next year's.

Choosing an approach for your size and situation

Not every institution should implement this the same way. The right answer scales with branch count, data maturity, and regulatory complexity.

Single branch, no workforce software. Do this in a spreadsheet. Six rows, one per operating day. Pull contribution by weekday from your finance reports, divide by the target, place shifts against your own observed hourly pattern. A free single-location scheduling tool is enough to publish and manage swaps. Do not buy enterprise software to solve a six-row problem.

Three to fifteen branches, community bank or credit union. This is where per-location pricing beats per-user pricing decisively, especially with large part-time teller rosters. You need one shared contribution table, one agreed per-person target across all branches, and one person who owns the monthly re-pull. The comparison you care about is per-location versus per-user cost at your actual headcount — with twenty part-timers across four branches, per-location typically wins by a wide margin.

Fifteen to fifty branches, multi-market. Demand-based scheduling earns its keep here: feed the tool a daily forecast and let it propose coverage, then override with your gross-profit counts. Compliance starts to matter — break rules, overtime thresholds, and predictive-scheduling ordinances differ by jurisdiction, and manual tracking stops being viable.

Fifty-plus branches, multi-state, possibly unionized. Enterprise workforce management is the infrastructure answer. Budget a multi-month implementation and a dedicated project owner. Understand what you're buying: forecasting, compliance, shift bidding, and integration — not the staffing philosophy. You still bring the gross-profit math; the platform automates its execution across a footprint too large to hand-manage.

The through-line across every size: the tool is the execution layer, never the decision layer. Software distributes, notifies, tracks, and enforces compliance. It does not know what a Friday at your Midtown branch is worth. You supply that, and the schedule follows.

Adjacent effects worth planning for

Changing the schedule changes more than the labor line, and the second-order effects are where implementations either compound or unravel.

Referral volume moves with coverage. A teller with a three-person line does not start a conversation about a CD rate or a mortgage refinance. A teller with breathing room does. That's the mechanism behind the "$400 doing average work, then dig for the next $400" framing — it only functions if the schedule leaves room for the digging. If you staff to the absolute minimum, you will hit your labor target and quietly kill your referral pipeline, and the damage shows up two quarters later in origination volume nobody traces back to a schedule change.

Cash logistics follow the peaks. Heavier Friday coverage means heavier Friday cash flow through the drawers, which affects vault limits, armored-carrier pickup timing, and end-of-day balancing. Tell your cash operations team before you re-cut, not after.

Cross-training becomes the flexibility lever. The single highest-leverage investment alongside this method is training platform bankers to run a teller window and tellers to handle basic account servicing. A cross-trained roster lets you schedule closer to the true count because you have real overflow capacity. Without it, you have to pad every day against the worst hour.

Digital channel shift changes the inputs. Every percentage point of transaction volume that moves to mobile deposit or ACH lowers window demand while barely touching advisory demand. That doesn't mean cut proportionally — it means the mix inside your contribution number is changing, and your per-person target may need to rise as the remaining work concentrates in higher-value interactions. Re-examine the target annually, not just the counts monthly.

The method travels. This is the same division that works in a department store, a wine shop, or an electronics floor — daily contribution divided by a per-person daily target, then shifts placed against the real hourly curve. What changes is the denominator's definition and the shape of the demand curve. A bank branch peaks at open, lunch, and payday Friday; a wine shop peaks Thursday evening through Saturday. The arithmetic is identical.

Staff reaction is a leading indicator. When you publish the first math-derived schedule, listen carefully to the complaints. "Tuesday feels dead now" usually means the math is right and the habit is loud. "Friday is unmanageable" usually means your target is too high or your hourly placement is wrong. The people at the windows will tell you which one it is faster than any report.

Related questions

How do I calculate a branch's daily gross profit if finance doesn't report it that way?

Ask finance for branch-level contribution or direct margin: net interest margin on the branch's deposits and loans, plus fee and service income, less cost of funds and direct expense. Most core banking systems support this cut. If it truly doesn't exist, a value-weighted transaction proxy works as a starting substitute.

Should Saturday be scheduled with the same method?

Yes, but expect a small count and let your operational floor govern. Saturdays typically produce well below weekday contribution while still requiring dual control and cash security, so the math often computes to two or three and the floor rounds it up. Judge Saturday on customer retention, not on labor efficiency.

How long before I see savings?

Expect noise for the first two weeks as staff adjust and swaps settle. A quarter is the honest measurement window for both labor cost and wait times. Judging in week two mistakes transition friction for a failed method.

Does this work for a branch with irregular, unpredictable traffic?

Yes, but flexibility replaces precision. Schedule a core team against your predictable contribution and add cross-trained floats or on-call part-time coverage for the unpredictable portion. Irregular traffic argues for more cross-training, not for abandoning the count.

What if the math says fewer people than I currently employ?

You don't have to cut. Reallocate first — move hours to peak windows, expand advisory and outbound calling capacity in the slow blocks, or cover a sister branch. Many institutions absorb the difference through natural attrition rather than reductions.

FAQ

What's the biggest mistake banks make when scheduling tellers?

Relying on habit or last year's grid instead of actual data. "That's how many we've always run" sets a flat line across a curved demand pattern, which guarantees idle staff on slow days and long lines at peak — in the same week, at the same branch. Match hours to demand and value, not to precedent.

How do I figure out the right number of tellers for my branch?

Two complementary methods. Divide the branch's average daily gross profit by an agreed per-person daily target to get the count. Then check it against a queueing view — track arrival times and transaction volumes for two to four weeks, identify peak windows, and confirm your peak coverage holds wait times under about five minutes. The division sets the number; the queueing view validates the placement.

Is it better to add tellers or speed up transactions?

Both help, and service speed often compounds better. Cutting thirty seconds off an average transaction through training or better technology can remove the need for a full position at peak. But complex work — business deposits, disputes, loan payments — resists compression, and forcing speed there degrades service. Speed up routine work; staff for complex work.

What's a realistic teller utilization target?

Roughly 60 to 75 percent during the busiest two-hour window. Below 50 percent means paid idle capacity. Above 85 percent means queue times climb sharply and burnout follows, because wait times rise non-linearly as utilization approaches full. Always measure during the peak window rather than across the whole day.

How much should I vary staffing across days of the week?

Let the contribution table decide rather than a fixed percentage. In practice, peak days commonly run meaningfully above the branch's own weekday average and slow days below it — a $3,600 Friday and a $2,000 Wednesday at the same $400 target is a nine-versus-five split. Pay cycles, the first and fifteenth, holidays, and local employer schedules can shift patterns further, so re-pull the table monthly.

Do I need scheduling software to do this?

No. A single branch can run the entire method in a spreadsheet with six rows. Software becomes worthwhile at three or more locations, where publishing, swaps, time tracking, and labor-law compliance stop being manageable by hand. Buy the execution layer once the coordination cost exceeds the license cost — never to supply the decision itself.

Sources

• https://www.bls.gov/ooh/office-and-administrative-support/tellers.htm
• https://www.federalreserve.gov/publications/files/2022-report-economic-well-being-us-households-202305.pdf
• https://www.aba.com/
• https://www.fdic.gov/analysis/quarterly-banking-profile/
• https://www.mckinsey.com/industries/financial-services/our-insights
• https://www.deloitte.com/us/en/insights/industry/financial-services.html
• https://www.ncua.gov/
• https://www.consumerfinance.gov/data-research/

Related on PULSE

• How Do I Get My Bank Branch Staff to Cross-Sell Accounts and Services?
• How Do I Get My Bank Staff to Cross-Sell the Full Product Set?
• How Many Employees Should I Schedule Each Day at My Department Store?
• How Many Salespeople Should I Schedule Each Day at My Electronics Store?
• How Many Employees Should I Schedule Each Day at My Toy Store?
• How Many Employees Should I Schedule Each Day at My Wine Shop?