What does the discount monitor show?
Every discount given in the range, drilled from the company total down through location, employee, transaction and item. The total is the realised discount, meaning the amount actually taken off the lines, including stacked codes, price rules and manual reductions, which is the figure that reconciles to the net-sales bridge.
It is split into automatic discounts, which follow the rules the brand configured, and staff-entered manual ones, which follow a person's judgement at the till or in an order edit. Both are legitimate; they are simply different to manage.
- The realised discount for the period, reconciled to the net-sales bridge.
- The split between automatic rule-driven discounts and staff-entered manual ones.
- The drill-down: company, location, employee, transaction, item.
- Policy alerts for oversized, over-limit, stacked and employee-outlier discounts against your own rules.
- Per-code before-versus-during verdicts, with the comparison window stated.
Why the headline rate is not the number
A code says fifteen off. A customer stacks it with a returning-customer rule and a bundle price, and the realised discount on the line is much larger. Averaged across a month, the brand's real discount rate can sit well above the rate printed on any code, and no single report in a store admin shows it.
The commerce engine computes realised discount per line and rolls it up, so the owner sees what was actually given. It is the difference between a policy and its outcome.
How do policy alerts follow your own rules?
The brand sets the rules: a maximum manual discount, a limit per employee per day, whether codes may stack, and the threshold for an outlier. The engine raises an alert when a rule is crossed: an oversized discount on a single item, a total over the limit, a stack that should not have been possible, or an employee whose pattern sits far outside their location's.
An alert is information, not an accusation. It names the transaction and the item so a manager can look at the actual sale before drawing a conclusion, which is how a loss-prevention process stays fair.
Per-code before-versus-during verdicts
For each code, the engine compares the period before it ran with the period during, using the brand's own sales, and reports whether the code appears to have driven incremental purchase or mostly discounted demand that was already there. The verdict shows the arithmetic, including the assumption it made about the comparison window.
This is where the largest recoveries usually sit. A code that runs permanently and mostly serves existing demand is a standing margin cost, and the verdict makes it visible enough to argue about with evidence.
What loss prevention here does not claim
It does not identify fraud. It identifies patterns that cross the brand's own policy, which a manager investigates. It does not publish an industry figure for discount leakage, because the engine reports the brand's own numbers only. And it cannot see a discount given outside the systems it reads, such as cash taken off the books.
It also never acts. Pausing a code, changing a limit or speaking to an employee are human actions; the engine ranks the evidence and drafts the message.
Which role owns it on ftes.ai?
The Pricing Officer FTE owns realised discount and is judged on it. It proposes pausing codes with poor verdicts, tightening limits where alerts cluster, and changing rules that let stacks through. The Store Officer FTE reads the per-location view for its own shops.
A brand that connects the store sees realised discount and the drill-down immediately, because it is computed from order and discount lines. Setting the policy rules takes a few minutes and turns the surface from a report into a monitor.
How does a brand set its own discount policy here?
The rules are the brand's, entered once: the maximum manual discount an employee may give, the daily limit per person, whether codes may stack, and the threshold at which an employee's pattern counts as an outlier. The engine does not impose a policy; it enforces the one the brand wrote.
Because the rules are explicit, an alert is never a matter of opinion. It names the rule that was crossed, the transaction that crossed it and the item it was given on, which is what makes the conversation with a store manager a conversation about a fact.