When the Till Doesn't Balance: Cash Control for Ghanaian Shops

It is a few minutes past closing. The cash in the drawer is GHS 60 less than the system says it should be.

Nobody knows why. The cashier does not know why. Tomorrow it might be GHS 15 over. The usual outcome is that the owner absorbs it, feels uneasy, and says nothing — which is the worst of the available options, because it teaches everyone that nothing happens.

A till that never balances is not necessarily a theft problem. It is almost always a measurement problem first, and until the measurement is right you cannot tell the two apart.

What Balancing Actually Means

The arithmetic is simple, and worth writing down because most disputes come from people using different versions of it:

Expected cash = opening float + cash sales − cash paid out − cash refunds

Then: counted cash − expected cash = the difference.

Three things break this in practice, and they are the reason most tills "never balance":

• The opening float was never counted. If you do not know what was in the drawer at the start, every closing figure is an opinion. Count it before trading, every shift, without exception.
• Cash was taken out during the day — a supplier paid, fuel bought, a staff advance — and never recorded. This is the single most common cause of an apparent shortfall, and it is not dishonesty, it is a missing note.
• Mobile money was recorded as cash. More on this below, because it now dominates.

Mobile Money Is Why Modern Tills Look Short

In many Ghanaian shops MoMo is now a large share of takings, and it has quietly broken the way cash was traditionally reconciled.

Money received by MTN MoMo, Telecel Cash or AirtelTigo Money is not in the drawer. If a cashier rings it up as "cash" because that is the fastest button, the system expects physical notes that were never there. The till appears short by exactly the MoMo total, every single day, and no amount of counting will resolve it.

Two habits fix this permanently:

• A separate button per payment method, used properly — cash, MoMo, card, credit — with split payments recorded as split payments when a customer pays partly each way.
• Reconcile each method against its own record. Physical cash against the drawer. MoMo against the merchant statement or the phone. Card against the terminal. One combined figure hides everything.

Once methods are separated, a genuine cash shortfall becomes visible again — and it becomes a much smaller, much more meaningful number.

Reconcile Per Shift, Not Per Day

This is the change that turns a vague worry into something you can act on.

"The shop was GHS 90 short today" is nearly useless if three people worked the till. You cannot tell whether it was one incident or thirty small ones, whether it happened during the lunch rush or at 8am, or whose shift it fell in.

Per-shift reconciliation gives you one row per cashier per session: opening float, cash sales, MoMo, card, cash paid out, counted cash, difference, open and close times, and the name. In SellarPro this is Reports → Register Report, and it is the first place to look when the numbers feel wrong.

With a few weeks of shift rows, patterns appear that a daily total would never have shown.

Reading the pattern

• Randomly over and short, small amounts — normal for a busy cash business. Leave it alone.
• Consistently short, never over — this is the one to look at. Honest mistakes scatter in both directions; a one-directional pattern rarely does.
• Short only on one person's shifts — worth investigating, but check the boring explanations first: are they always on the busiest shift, do they handle most of the MoMo, were they trained on refunds?
• Short only at particular times — usually a process problem. Rushes cause miscounting, and shift handovers with no count in between make it impossible to attribute anything.
• A gradual drift over months — often an uncounted float that has slowly gone wrong rather than any single event.

Where the Money Actually Goes

In order of how often each turns out to be the answer:

1. Cash paid out and never recorded. Fuel, a supplier, transport, an advance. Fix: a paid-out entry with a reason, every time, no exceptions.
2. Payment method recorded wrongly. Almost always MoMo as cash.
3. Change given incorrectly during a rush. Genuine, unavoidable, and small. This is what a variance threshold exists to absorb.
4. Refunds handled outside the system. Cash out of the drawer with no transaction behind it — the till is short and the stock count is wrong too.
5. Discounts given without authority. The sale records full price, the drawer holds less. Whether a cashier can discount at all should be a role setting, not a matter of trust.
6. The opening float was wrong. Everything downstream inherits it.
7. Deliberate theft. It happens, and it is last on this list because it is far less common than the six above — and because accusing someone before eliminating those six is how good staff leave.

A Routine That Takes Five Minutes

• Opening: count the float and record it. Two minutes.
• During the shift: every cash removal recorded with a reason at the moment it happens.
• Handover: close one shift and open the next with a count in between. Never let two people share a session — it makes attribution impossible and is unfair to both.
• Closing: count cash, check MoMo against the phone, check card against the terminal, record the difference.
• Weekly: read the shift rows together, not one at a time. The pattern is the point.

Set a variance threshold and stick to it. Anything inside it is noise. Anything outside it gets a reason recorded against it. What must not happen is a difference being quietly absorbed — that is how a small recurring loss becomes permanent, and how the one shift that genuinely mattered gets lost among the ones that did not.

The Controls That Prevent Most of This

• Individual logins. A shared login means no reconciliation is ever attributable. This is the foundation — nothing else works without it.
• Role-based permissions. Who may discount, void, refund, or open the drawer without a sale. Most owners give cashiers the till and nothing else. See POS system.
• Manager authorisation on voids and refunds. The two transactions most often used to cover a shortfall.
• An activity log. Every deleted sale or edited price carrying a name and a timestamp changes behaviour long before anyone has to look at it.
• Per-method totals at close. Non-negotiable in a MoMo economy.

Related reading: preventing stock theft in retail shops covers the stock side of the same problem, and why your POS profit figures are wrong explains a different number that misleads for a different reason. For the daily reporting picture, see reports and analytics and daily sales reporting.

A till that balances is not the goal in itself. The goal is knowing, quickly and without an argument, whether today was normal.

Frequently Asked Questions

Part of our Reporting & Analytics resources — see the full guide and related tools.