A Discount With No Name on It
A discount at the till is an amount and a sentence. There is no approver, no reason code, no percentage and no person on the record. Meanwhile a stock write-off of the same value needs a dedicated permission and a second pair of eyes that no permission can override.
A cashier can give away value in two ways. They can write off stock, or they can discount a sale. Both reduce what the business ends up with by the same amount. Only one of them is controlled.
What a discount record contains
Four fields. The sale it belongs to, optionally the specific line, a description, and an amount.
That is the entire record. No user. No approver. No reason from a list. Not even a percentage — the discount is stored as a figure, so a report cannot tell a ten per cent courtesy from a ten per cent price error without knowing what the line was worth.
There is attribution, and it is worth stating precisely because it is better than nothing and weaker than it sounds: the discount belongs to a sale, and the sale belongs to a cashier and a session. So you can find out who was at the till. What you cannot find out is who decided.
The record knows which sale gave money away. It does not know who authorised it, or why.
The comparison that makes the case
Take the same amount and remove it from the business two different ways.
| Write off stock | Discount a sale | |
|---|---|---|
| Value threshold | Yes, set per organisation | None |
| Dedicated permission above it | Yes, its own grant | None |
| Second person required | Yes, and no grant overrides it | None |
| Reason | From a maintained list | Free text |
| Who did it | Recorded on the adjustment | Inferred from the sale |
| Net effect on the business | The same | The same |
The last row is the point. These are the same economic act performed through two different doors, and the doors were built by different people at different times with different levels of care.
It is worth being fair about why. A write-off is unusual and interruptible: somebody is at a desk, and a second approval costs a few minutes. A discount happens with a customer standing at the counter, and a control that stops a queue is a control that gets removed in week two. The right answer at the till is almost certainly not the same as the right answer in the stockroom.
But "not the same" is not the same as "none".
Why a small remote retailer feels this most
Not because of dishonesty. Because of the opposite — because in a small community shop the cashier knows every customer, and courtesy discounting is a normal, well-intentioned, socially expected part of the job.
Which means the discount total is not a fraud figure. It is a management figure, and it is one of the most useful numbers a small retailer can have: how much did we give away this month, on what, and at which counter?
With an amount, a sentence and no structure, that question is answerable only by reading. And nobody reads.
-
Agree a fixed vocabulary for the description field
Four or five words, written on a card at each till: DAMAGED, PRICE-MATCH, STAFF, GOODWILL, ERROR. Free text is only unsearchable when everybody writes something different, and this costs nothing.
-
Total discounts by session, monthly
The discount belongs to a sale and the sale belongs to a session and a cashier. That chain is enough to produce a per-counter figure in a spreadsheet, and the figure is the whole point.
-
Set an expectation, not a limit
Since the system will not enforce a ceiling, publish one. "Anything over this amount, come and find me" is a control implemented in people, and in a small shop it works better than a screen would.
-
Watch the ratio, not the total
Discounts as a percentage of sales, per counter. A busy till will always discount more in absolute terms, and comparing raw totals will point you at your best cashier.
Three fields, in ascending order of cost
This is one of the cheaper items we publish about, and the ordering matters more than the list — the first one alone would close most of the management gap.
A reason from a maintained list
The largest gain for the smallest change. A coded reason makes every discount countable, and the reason list already exists as a pattern elsewhere in the product.
The person who authorised it
Distinct from the cashier. In a small shop these are often the same person and the field is still worth having, because the day they differ is the day you want to know.
A threshold with a supervisor override
The most intrusive and the one to think hardest about. A control that stops a queue will be worked around; a control that fires rarely will not.
No dates on a public page. If discount leakage is a live problem rather than a theoretical one, tell us what your counters look like and we will scope it in writing.
Scope till controlsThree questions about discounting at a till
Who authorised this discount?
A good answer sounds like
A named person on the record.
What it actually means
Ours infers the cashier from the sale and records no authoriser. Ask to see the record, not the receipt.
Show me discounts by reason for last month.
A good answer sounds like
A grouped total.
What it actually means
Needs a coded reason. With free text you get a list to read, and nobody reads it.
Can a cashier discount without limit?
A good answer sounds like
A threshold, or an honest no.
What it actually means
Ours is no limit. Whether that matters depends on your counter, and it should be your decision rather than a discovery.
Our position
Discounting at the till is attributable and not controlled, and the same value taken out through a stock write-off is controlled properly. Close the gap with people rather than waiting for a feature: a fixed vocabulary in the description, a monthly total by counter, a published expectation and a look at the ratio rather than the total. All four are free, and together they turn an unexamined number into a management one.
What AWRA OpsHub does today
- Discounts recorded against a sale and optionally a specific line, with a description and an amount.
- The sale carrying its cashier, counter and cash session, so a discount is indirectly attributable to a person and a shift.
- Cash sessions with a float, drops, expected and counted cash, and a stored variance.
- Returns at the till reversing both revenue and cost of sales.
- Pricing anomaly detection at the item level — below cost, thin margin, zero price.
What it does not do
- Any approver, authoriser or user on a discount record.
- A coded reason. The description is free text.
- A percentage. The discount is stored as an amount.
- Any threshold, limit or supervisor override on discounting.
- Any report grouping discounts by reason, cashier or counter.
Not ours, by choice
- The right control at a till is genuinely not the right control in a stockroom, and we would resist copying the write-off approval across without thinking about the queue.
- The attribution chain through the sale to the session is real and usable in a spreadsheet today. It is weaker than a field and stronger than nothing.
- Nothing here is Tongan. It is what an uncoded discount does to a management figure; a small high-trust shop is where courtesy discounting is most routine.
Put five words on a card at each till
The cheapest control on this page. A fixed discount vocabulary turns free text into something you can count, and it costs one printed card.
Set up the vocabularyFrequently asked questions
Can I stop cashiers discounting entirely?
Not through a setting. There is no threshold and no permission gating a till discount, so the control has to be a published expectation and a monthly review. In a small shop that is usually more effective than a screen anyway, because everybody knows the figure is being looked at.
Does a discount affect the recorded margin?
Yes — the sale is recorded at what was actually charged, so margin analytics reflect the discounted figure. What you cannot do is separate out how much of a thin margin was discounting rather than pricing, because the discount reason is free text.
Why is a stock write-off controlled and this is not?
They were built at different times for different problems, and nobody compared them. We publish the comparison because it is the fastest way to see where a product's controls are uneven — and that unevenness is normal in every system, not unique to this one.
