Independent verification of public-sector overtime pay
One closed pay period, recomputed from your own collective bargaining agreement and the FLSA baseline, then compared against what your payroll system actually paid.
Twenty-one published cases. None found by a routine check.
We studied 21 published cases across 19 public employers — federal court filings, published audits and official reports. In every case where the record says how the error was found, it was found by an employee, a union, a plaintiff or an auditor. None was found by a routine control that recomputed pay.
None of the 21 records describes the error being found by an existing routine control that recomputed pay independently. Minnesota's payroll system flagged 97% of records for review; the agencies approved them anyway.
These errors do not announce themselves
A pay-rule error does not produce a missing number. It produces a number that is wrong and looks right, on a payslip that balances, in a system that reports no exception. It survives reconciliation because reconciliation checks that the payroll system agrees with itself.
We studied 21 published cases across 19 public employers — federal court filings, published audits and official reports. In every case where the record says how the error was found, it was found by an employee, a union, a plaintiff or an auditor. None was found by a routine control that recomputed pay.
Some ran for years. In one county, more than 200 employees were affected and one paramedic was underpaid by $9,000.
Three windows
A new agreement.
Rules change and the system does not. This is the most common trigger in the records we studied.
A system migration.
Errors appear within weeks of go-live and can persist for years.
A changed pay code.
A single code updated for one purpose changes overtime for another.
Independent by construction
A check that reads the payroll system's own output is a restatement of that output. It cannot disagree with the system it is reading.
So we do not take the answer from the payroll system. We apply your collective bargaining agreement and the FLSA baseline from first principles, rebuild each employee's work period from hours actually worked, and compare the result against what was paid. Differences are reported in both directions, as underpayments and as overpayments.
The usual check
Independent recomputation
The computation is deterministic. The same inputs and the same rule version always produce the same result. There is no AI anywhere in the calculation path. It carries 145 automated tests that run on every change, and we will walk you through any of them, including the tests themselves.
Where the inputs do not support an answer, that employee is reported as unresolved rather than given a best guess.
What the work covers, and what it does not
What a lookback covers
- Work period and threshold: whether overtime was derived from the right hours over the right period
- Regular rate under 29 U.S.C. 207(e): whether the premium was paid on the right rate, with differentials, longevity, certification and incentive pay in the base
- Completeness: whether everyone who should have been paid appears in the run at all
- Retroactive pay, where you supply the prior payments
- A written exception for every difference, with a clause reference and an audit trail
What it does not cover
- Whether recorded hours were actually worked
- Exemption classification, which is flagged and never decided. That determination is for your counsel
- Remediation, recalculation for payroll, or anything written back to your systems
- Legal advice of any kind
We never run payroll and never write to your systems. Output is advisory. You approve and you submit.
Four steps, one closed period
We read your agreement first.
Collective bargaining agreements are public. We read yours before we quote, so the scope reflects your rules rather than an average.
A short call about your data.
We establish what your extracts contain and which closed period is most useful to test. No charge, and nothing for IT to build.
A fixed fee, agreed up front.
Scoped from what the agreement and the data actually require. Never a share of what is found.
One closed pay period, recomputed.
You receive an evidence pack: every exception, with the clause it turns on.
We are never paid a share of what we find
A fee that depends on the answer gives us an interest in the answer. Roughly half of what these engagements find is overpayment rather than underpayment, and we will report that to you as plainly as the rest.
A clean result is also a result. It establishes that pay matched the rules for the period tested, which is worth knowing and worth being able to show.
Who does this work
Pay-Rule Assurance is a service of Ai Partner Solutions, a Florida firm. The work is led by Josh Riley.
The engine behind it was built out of an engagement with a Florida county government, verifying overtime calculations for emergency services personnel against a collective bargaining agreement and the FLSA.
We are a specialist practice rather than a large firm, which is the reason the method is deterministic and documented rather than resting on headcount. Every calculation is reproducible, every exception cites the clause it turns on, and the rules are versioned data rather than code. We will show you the engine, the rule pack and the tests.
Paid Correctly?
Our evidence paper examines 21 cases across 19 public employers: seven mechanisms by which public-sector overtime goes wrong, and why the usual checks do not find them. Every case carries its document number, court or office, date and source location. Thirteen are drawn from primary records.
It is available on request.
We will send the paper by email. We do not add requesters to a mailing list.
One closed pay period, one question
Does what you paid match what your agreement and the FLSA require?
A 30-minute call establishes which period would be most useful to test and what an extract would need to contain. No obligation and nothing to prepare.
Josh Riley · Ai Partner Solutions · josh@aipartnersolutions.io · (813) 308-9054