Payroll has a failure mode that's easy to miss until it hits you: cross-employee leakage.
It happens when one employee's rate, configuration, or hours silently affects another employee's pay. In a small team it might be a rounding annoyance. In a 50-employee restaurant or staffing operation it can mean wrong payments, wrong tax, and a lot of manual re-checks.
We didn't want to assume our engine was safe from this. So we built a test to try to break it.
The test
We set up multiple employees in the same test company with different hourly rates and different configurations, then ran payroll for all of them in the same calculation cycle.
The questions we asked:
- If Employee A's hourly rate is EUR 10.00, does it stay EUR 10.00 and not bleed into Employee B's calculation?
- If Employee B's configuration changes, does Employee A's pay stay exactly what it was?
- Does running payroll for one employee contaminate the calculation for another?
A payroll engine that isn't careful about state can fail these in subtle ways — shared caches, shared mutable objects, a calculation that accidentally reads the wrong employee record.
What we found
Zero leakage.
The test passed 3 out of 3 scenarios:
- Standard hourly baseline — regular weekday hours calculated correctly per employee under the test company.
- Split-shift scenario — a restaurant-style split shift with an unpaid break calculated correctly, with regular vs daily overtime split at the right threshold.
- Payroll isolation — modifying one employee's rate or configuration had no effect on any other employee's payroll.
None of the existing payroll tests regressed. The isolation held.
Why this matters for buying payroll software
If you're running payroll for multiple employees, you need two guarantees:
- Correctness per employee — each person's pay is right.
- Isolation between employees — one person's data can't quietly change another's.
Most vendors claim both. We wanted proof, not a claim. So we tested for leakage directly and published the result here.
If you're evaluating Chegatta for real payroll, this is the kind of test you should ask every vendor about: "show me you've tested for cross-employee leakage, not just that a single employee calculates correctly."
This is one piece of a bigger picture
Zero leakage is necessary, not sufficient. Payroll has other risks: wrong tax tables, wrong Social Security bases, overtime treatment that doesn't match your sector, locked runs that can be changed anyway.
We cover the bigger picture here: Why we publish our payroll test results instead of pretending everything is perfect.
And the specific scenario that breaks naive payroll systems here: Restaurant payroll is harder than office payroll — here's what we tested.