The days from each order to its payment, an unpaid one counted up to the last known date; a level by rules, from paid on time to final notice; and the orders to chase. On a synthetic sales dataset: one invented shop, 2024-2025.
Two house rules checked on every order, a zone each: no store or phone order on a weekend, when the counter is closed; and no two orders of one customer on the same day. On a synthetic sales dataset: one invented shop, 2024-2025.
Three leads side by side: the lines entered twice, the quantities absurd for their product, and the customers whose amounts swing far more than the others’. Each ends on the list of what to look at. On a synthetic sales dataset: one invented shop, 2024-2025.
What each order billed against what was paid for it, both ways: the orders never paid, the payments for no known order, the orders paid a different amount. On a synthetic sales dataset: one invented shop, 2024-2025.
The raw file checked against written rules, errors and warnings, with a report of what breaks; then the fixes, a second check, and a gate that lets the rest run only on a file without errors. On a synthetic sales dataset: one invented shop, 2024-2025.
