Cash handling is the subject most operators least want to turn into surveillance, and with good reason. A programme that feels like watching people will cost more in trust than it recovers in cash. It is also unnecessary, because the useful signal is not about people at all.
Short on time
Start with the transaction
Review drawer openings, voids and refunds in the context of a POS transaction.
Separate signal from accusation
An exception warrants review; it is not proof of employee wrongdoing.
Match the timing
The drawer record helps more when matched to the sale and the payment-area view.
Set branch-specific rules
Operating patterns differ by branch; inspect causes before changing thresholds.
The unit is the exception, not the person
A no-sale that opens the drawer without a transaction, a void after payment, a refund at an unusual hour, a drawer open with no order on screen — each of these is an event with a POS, a time and an amount attached. That is a complete record on its own. It does not need a name attached to be actionable.
Aggregate first, investigate second
Almost every individual exception has an innocent explanation, and investigating them one by one is both insulting and useless. The signal is in the concentration: one POS, one hour, one type of exception, repeating over weeks. That is a question worth asking, and it can be asked about a POS rather than about a person.
Ask the question about the POS. The shift record answers the rest, and it is yours already.
Keep the rota out of the system
There is no need to feed staff schedules into anything. If a pattern appears at a particular POS at a particular time, the branch’s own rota answers who was there, held by the people who are responsible for that decision. That separation is what keeps this a control rather than a monitoring programme, and it is worth protecting deliberately.
Setting the threshold
Start loose. A threshold tight enough to catch everything will bury the branch in alerts and be switched off within a fortnight. Two weeks of observation will tell you what normal looks like at each site, and normal differs more between branches than most groups expect.
ROI calculator
Four inputs, and the workings stay on screen.
Open the calculator


