The Complete Guide to Cash Handling and Till Close

Cash is the one part of a retail or restaurant day that never fully forgives a mistake. A card transaction that goes wrong leaves a record you can chase. A twenty that goes missing from a drawer just leaves a gap, and the gap shows up at close as a number nobody can fully explain. Most operators handle this the same way: someone counts the drawer at the end of the night, compares it to what the register says, writes the difference on a sheet, and moves on. It works, roughly, until the day it does not. A drawer is short by a meaningful amount, the manager cannot tell whether it was a miscount, a bad starting bank, a refund keyed to the wrong tender, or something worse, and there is no trail to follow.

This guide is our attempt to lay out the whole cash cycle in one place, from the moment a drawer is loaded in the morning to the moment a deposit clears at the bank. We build software for drawer counts and till reconciliation, so we spend a lot of time talking with owners and managers about where their process breaks. The patterns are remarkably consistent across coffee shops, quick-service restaurants, hardware stores, and boutiques. The same handful of gaps produce most of the pain. Each section below covers one theme, explains what good looks like, and points to the deeper articles where we go into the details. You can read it straight through or jump to the part that hurts most right now.

Start with the bank: why every close begins in the morning

Every reconciliation problem you will ever chase at closing time depends on one number that was set hours earlier: the starting cash bank. If the drawer opened with an amount that was never verified, then the expected total at close is built on sand. The register will happily tell you that cash sales were 412 dollars and 18 cents, and you will add that to what you believe the drawer started with, but if the morning count was 200 instead of the 150 written on the sheet, you will spend twenty minutes hunting a fifty-dollar overage that never existed. The starting bank is not a formality. It is the anchor for every calculation that follows, and it deserves the same care as the final count.

A good starting bank has three properties. It is a fixed, known amount that the whole team recognizes, so any deviation is obvious. It is counted by the person taking responsibility for the drawer, not by whoever loaded it, because the person who signs for the bank should be the one who verified it. And it is built from the right mix of denominations for your business, which is a question of what customers actually hand you. A store that sells a lot of low-ticket items for cash needs far more ones and quarters than a restaurant where most cash tickets land near a twenty. Running out of change mid-shift forces a bank adjustment, and every adjustment is one more chance for the paper to drift away from the drawer.

The operators who get this right treat the morning count as a mini close. The opener counts the drawer against the standard bank, records the result, notes any difference before a single sale rings, and either corrects the bank or flags it. That takes two or three minutes and removes an entire category of end-of-day mystery. It also creates a useful habit: staff learn that a drawer is always counted at a handoff, in both directions, and that nobody is ever holding a drawer they did not personally verify. Our article on setting the right starting cash bank goes deeper into denomination planning and how to size the bank for different volumes, and the guided count article shows how the same counting method works at open and at close.

Counting the drawer: method beats memory

Ask five cashiers to count the same drawer and you will often get three different totals. That is not because anyone is careless. It is because counting a mixed drawer in your head, under time pressure, at the end of a long shift, is genuinely hard. People lose their place in a stack of ones, forget whether they already added the rolled coin, or count a bundle of fives twice. The fix is not to tell people to concentrate harder. The fix is to give them a method that does not rely on concentration: count one denomination at a time, record each subtotal as you go, and let the arithmetic happen somewhere other than in your head.

A guided count walks the counter through the drawer in a fixed order, usually from largest bill to smallest coin, and asks for a quantity of each rather than a dollar amount. Twelve twenties, seven tens, nineteen fives, forty-three ones, then rolled coin, then loose coin. Entering quantities rather than dollar values removes multiplication mistakes, and it produces a record that is far more useful later. When a drawer is off, seeing that it had forty-three ones instead of the usual thirty tells you something. Seeing only a total tells you nothing. This is also the natural place for a blind count, where the person counting cannot see the expected total until they commit their count. Blind counts stop the very human tendency to nudge a number until it matches.

Speed matters too, because a slow close is a close that gets rushed on busy nights, and a rushed close is where errors and shortcuts creep in. The two are not in tension. A structured, denomination-by-denomination count is usually faster than an unstructured one, because nobody restarts halfway through. Standardizing the drawer layout, keeping coin rolled and labeled, and pulling large bills to the safe during the shift all shave minutes without touching accuracy. Our guided drawer count article covers the step-by-step procedure and how to handle recounts, and the article on speeding up the closing count lists the practical adjustments that make the biggest difference on a busy night.

End-of-day reconciliation: turning a count into an answer

A count is a number. A reconciliation is a comparison. The difference matters because a drawer total on its own does not tell you whether anything went wrong. To reconcile, you take what the drawer should contain, which is the starting bank plus cash sales minus cash refunds minus any paid-outs and safe drops, and you compare it to what the drawer actually contains. The gap between those two figures is the over or short for that drawer, and it is the single most important number produced by the close. Everything else in the process exists to make that number accurate and explainable.

The expected side of the comparison is where most reconciliations quietly go wrong. Managers pull a cash sales figure from the register report and forget that a paid-out for a delivery tip, a safe drop at three in the afternoon, or a refund that was rung as cash but handed back as a card credit all move the expected total. When those adjustments are captured as they happen, with a note and a signature, the expected figure is trustworthy. When they are reconstructed from memory at eleven at night, the reconciliation turns into an argument. The practical rule is that nothing leaves or enters a drawer during a shift without a record, and every record feeds the expected total automatically.

Reconciliation also needs a clear definition of done. A drawer is closed when the count is entered, the expected total is computed, the difference is recorded with a reason code if one applies, and the person responsible has signed off. Not before. Stores that let a close sit half-finished, with the count on a scrap of paper and the comparison done tomorrow, lose the ability to investigate while memories are fresh. The end-of-day reconciliation article walks through the full sequence and the adjustments people forget most often, and if you run more than one register, the multi-register close article explains how to sequence drawers so the last one is not a scramble.

Over and short: reading the pattern, not the day

A single drawer that is three dollars short tells you almost nothing. Someone probably handed back the wrong change once. A drawer that is short in the same range every Tuesday and Thursday for a month is a signal, and the difference between the two is tracking. Over and short only becomes useful when it is recorded consistently, by drawer and by cashier, and reviewed as a series rather than as isolated events. The goal of tracking is not to catch people. Most variances are honest mistakes, and honest mistakes have patterns too: a cashier who is always slightly over is probably shortchanging customers by accident, which is a training issue and a customer service issue before it is a cash issue.

The most useful view is a simple log with date, register, cashier, expected, counted, difference, and a short reason. Over time, that log answers questions you could never answer from a single close. Is one register consistently off, which points to a hardware or setup problem? Is one shift consistently off, which points to a training or supervision gap? Is one person consistently off, and in which direction? Netting overages against shortages across a week is the classic mistake here. A drawer that is ten over on Monday and ten short on Tuesday is not a drawer that balanced. It is a drawer that was wrong twice, and treating the net as zero hides both errors.

Tolerance thresholds turn the log into action. Decide in advance what size of variance is normal noise, what size triggers a recount, and what size triggers a conversation with the cashier and a note in their file. Publishing those thresholds to the team removes the feeling of arbitrary scrutiny and makes it clear that the process applies to everyone. Our article on tracking over and short by cashier goes into how to set thresholds and read the patterns, and the article on reducing shrink through tighter cash handling connects those patterns to the broader loss picture in a store or restaurant.

Deposits and the safe: closing the loop with the bank

The drawer balancing at close is only half of the cash cycle. The other half is what happens to the cash after it leaves the drawer: it goes into a safe, gets bundled into a deposit, travels to the bank, and eventually appears on a statement. Every handoff in that chain is a point where money can go missing, be miscounted, or simply be forgotten. A store can have perfect drawer reconciliation and still lose money between the safe and the bank because nobody compared the deposit slip to the sum of the drawers, or because a deposit sat in the safe for four days and nobody noticed one bag was gone.

A deposit log fixes this by recording, for every bank drop, the date, the amount, which closes it covers, who prepared it, who carried it, and the bank confirmation when it comes back. The log should tie back to drawer closes so that the sum of drawer cash for a given day equals the deposit for that day, adjusted for whatever was retained as the next day's banks and change fund. When that relationship holds every day, a bank discrepancy is easy to isolate. When it does not, a discrepancy sends someone digging through a week of register tapes. The deposit log is also the document a bookkeeper or accountant will ask for first when reconciling the bank statement, so keeping it accurate saves real hours on the accounting side.

Safe management belongs in the same conversation. Mid-shift drops from the drawer into the safe reduce the amount at risk in the register, but each drop must be recorded and must reduce the expected drawer total, or the close will show a phantom shortage. Regular safe counts, ideally at every manager handoff, catch problems before they compound. Our deposit log article explains how to structure the log and reconcile it to the bank, and the audit trail article shows how drawer closes, safe drops, and deposits chain together into a record that stands up to scrutiny.

The audit trail: protecting the business and the honest cashier

An audit trail sounds like something only an accountant cares about, until the first time you need one. A customer disputes a refund. A cashier says the drawer was already short when they took it over. An owner notices deposits have been running a little light and wants to know since when. In each case, the question is the same: who did what, to which drawer, at what time, and what did the numbers say at that moment. If the answer lives on a clipboard with handwriting that could be anyone's, you do not have an audit trail. You have a story that someone can dispute.

A real trail has a few non-negotiable properties. Every count is attributed to a specific person and a specific time. Every adjustment to the expected total, whether a paid-out, a drop, or a correction, has a reason and an author. Records cannot be silently overwritten; a corrected count is a new entry that references the old one, not a replacement. And the trail is reviewed, because a record nobody reads is not a control. This matters as much for honest staff as for the business. When a cashier can show that they counted the drawer at handoff and it was already short, the trail protects them. Without it, suspicion falls on whoever happened to be holding the drawer last.

Shrink is the wider frame here. Cash shrink in a store or restaurant is rarely one big theft. It is usually a mix of small errors, sloppy refunds, unrecorded paid-outs, and occasionally a person who has noticed that nobody is looking. Tight cash handling shrinks all four at once, because the same controls that catch mistakes also make dishonesty visible early. The audit trail article covers what to record and how to review it, the shrink article puts cash controls in the context of overall loss prevention, and the over and short tracking article explains how a consistent log turns the trail into something you can actually read.

Scaling the close: multiple registers, shifts, and locations

Everything above gets harder in proportion to the number of drawers. One register with one closer is a manageable ritual. Six registers, two shift changes, a bar drawer, and a drive-thru drawer is a logistics problem, and the failure mode is predictable: the closing manager runs out of time, closes the easy drawers carefully, and rushes the last two. The next morning, the opener inherits a bank that was never confirmed, and the cycle of unexplained variance starts again. Scaling the close is mostly about sequencing, standardization, and distributing the work so that no single person is counting everything at the worst possible hour.

Sequencing means closing drawers as they go idle rather than all at once. A register that stops taking sales at eight can be counted at eight, while the floor is still staffed, instead of waiting until the last customer leaves. Standardization means every drawer has the same bank, the same layout, and the same count procedure, so any trained staff member can close any register without asking how this one is different. Distribution means cashiers count their own drawers, blind, and a manager verifies, rather than a manager counting every drawer alone. That splits the work and adds a second set of eyes at the same time.

Multi-location operators face one more layer: consistency across stores that never see each other. The same bank, the same procedure, and the same reason codes everywhere make it possible to compare locations honestly and to move a manager between sites without retraining them on cash. It also makes the numbers meaningful at the owner level, because a variance report only helps if every store defines variance the same way. Our multi-register close article walks through the sequencing in detail, the speed article covers the small operational changes that add up on busy nights, and the starting bank article explains how to standardize banks across drawers that see very different traffic.

More guides on this topic

Further reading from the TillClosr blog, each answering one specific question in depth.

Good cash handling is not complicated, but it is unforgiving of gaps. A verified starting bank, a guided count with quantities by denomination, a reconciliation that captures every adjustment as it happens, a variance log read as a series, a deposit log that ties to the bank, and an audit trail that names people and times: those six pieces together turn the close from a nightly source of anxiety into a routine that takes minutes and produces numbers you can trust. You do not need software to do any of it. A well-designed sheet and a disciplined team will get most of the way there. Where software helps is in removing the arithmetic, enforcing the sequence, and keeping the trail without anyone having to remember to keep it. Start with whichever gap costs you the most sleep, fix that one thoroughly, and move to the next.

If you are not sure where to start, count your drawers tomorrow morning against the bank they were supposed to have. If they match, your close is probably in decent shape and the articles on variance tracking and deposits will sharpen it. If they do not, begin with the starting bank and the guided count, because nothing downstream can be trusted until those two are solid.