Case Study
Payout Math: incentive payouts, re-derived from the program’s own rules
The rules are published. The question is whether the payouts followed them.
Figures on this page are from the live tool; the data is synthetic and seed-locked.
The number nobody re-derives
An incentive program pays against written rules: tiers, windows, eligible products, one bonus per serial. Then real claims arrive in volume, approvals happen under deadline pressure, and finance books what the system says. If your program is like most, no one has ever re-derived its payouts from its own rules. Payout Math is what that looks like when someone does.
What Payout Math does
The verdict. One program year of an invented HVAC manufacturer: $17,597 of $521,804 approved for payout should never have been approved, and $7,336 was earned under the program’s own rules and never paid. The two numbers are never netted: an underpaid contractor is not made whole by someone else’s overpayment.
The claims. All 129 wrong decisions, each with the rule it breaks, the evidence, and the dollars, in one filterable table. Every line can be reproduced by hand from the published rulebook; the tool asserts nothing it cannot show.
The program. The view a sponsor forwards: 61.2% of wrong decisions started with something a member submitted, but member errors averaged $98 while the sponsor’s own calculation errors averaged $344, and $13,871 of the total sits on just 10 distributor rebate statements. The members are not where the money leaks. The spreadsheet is.
How the proof is possible
Galewick Climate Systems is invented, and the invention is disciplined: every parameter in its rulebook is a seeded draw from a research-plausible range, chosen by no one, so nothing can mirror a real program. The kinds of error come from nineteen years of running incentive operations, stated as professional judgment, never dressed up as a citation. The validator reads only what the sponsor’s systems would hold and finds all 129 planted defects exactly, and nothing else. Two published benchmarks frame the result, both named on the tool page: a rebate-audit vendor’s 6 to 15% range for claims carrying a defect, which this data sits inside, and an older Forrester figure for overpaid share, which it sits below. The error rates were not adjusted to close the gap.
What this looks like on your program
Your rules are already written. The engagement runs this same validation on your claims and statement exports: every payout re-derived, every exception shown with its rule and its dollars, both directions, never netted. A program does not need to be large to leak, and the demo program here is deliberately the size of a real one.
Data is synthetic; methodology and deliverables are real.
Start in writing.
A few minutes by form, no call. Describe the program and the exports you have, and I will tell you what a re-derivation would check.