Most payroll systems serving California crews carry an earning code for the meal and rest premium, set up correctly on day one. Ask how many times it has been used. A code with zero uses beside a year of late meal punches is not a clean record. It is a record of exceptions that never reached payroll, and the reason they never reached it is the finding.
Open the earning-code list in almost any payroll system that runs a California hourly workforce and there it is. MEAL PREM, or BREAK PREMIUM, or 226.7: one hour, paid at the regular rate, created when the system was implemented because the implementation checklist asked for it.
Now pull the usage report for that code. Twelve months. There are two kinds of answer. One is a number in the dozens or hundreds, which means the operation catches exceptions and pays them. The other is zero.
Zero is where this post lives. A premium code that exists and has never fired is a system that is compliant on paper, because the mechanism for paying the premium is present, and failed in operation, because nothing feeds it. The failure is structural. It is built into how the exception travels, or does not travel, from the time clock to the check.
Labor Code §226.7(c), the operative sentence:
If an employer fails to provide an employee a meal or rest or recovery period in accordance with a state law ... the employer shall pay the employee one additional hour of pay at the employee's regular rate of compensation for each workday that the meal or rest or recovery period is not provided.
The construction wage order repeats it for meal periods in §10(F) and for rest periods in §11(D), in the same words: one hour of pay at the regular rate of compensation for each workday the period is not provided. Ferra v. Loews Hollywood Hotel (Cal. 2021) settled that the rate is the regular rate of compensation, not the base hourly rate, so the code has to pull in nondiscretionary bonuses and differentials the way overtime does.
That is the whole obligation. It is small, per day, and it is triggered by an event the time system already records. Which is why a code that never fires is so specific a symptom: the event is happening, or it is not, and either way the record should say which.
An exception has to travel from a punch to a payroll line. There are five places it dies on the way, and an operation with a zero-use code has at least one of them.
The auto-deduct. The time system subtracts thirty minutes from every shift over a set length, whether or not a meal was punched. The record now shows a meal period on every shift. There is no exception to route, because the deduction erased it before anyone could see it. On the days the meal was not taken, the record is wrong in the employer's favor, and the way that gets proved is the driver's telematics, the gate log, the foreman's text at 12:40 asking the crew to finish the pour. After that, the question in the room is what else in the record is wrong the same way.
The report nobody owns. The time system can produce a meal-exception report, and often it does, on a schedule, to an inbox. Reading it is nobody's job. Or it is read, and the reading ends there, because there is no step that turns a line on the report into a line on the check. The report is a record that the employer knew, every pay period, and did nothing with the knowledge.
The approval that absorbs the flag. Timecards go to a supervisor before payroll. The supervisor sees the missing meal punch and does what supervisors do with an error: fixes it. A meal punch is added at 11:30 and 12:00, from memory or from what the day was supposed to look like, and the timecard is approved. The exception is gone, and so is any trace of what the record said before it was corrected. We have written about why a correction has to sit beside the original; this is the case where the correction is the whole problem, because it converts a flagged shift into a compliant one with nobody's name on the change.
The mapping that was never built. The code exists in payroll. The rule that would put one hour on it when the time system flags a shift was never configured, because the two systems were implemented by two vendors in two quarters and nobody owned the seam. The checklist asked whether the code existed. It did not ask whether anything fed it.
The break that generates no punch. Rest periods are paid and unclocked, so there is no time record to flag. The only way a rest-period exception exists is if an employee reports it or attests to it. Without a reporting channel and a log, the rest-premium code has nothing to fire on, structurally, forever. Its zero is not evidence of compliance. It is evidence that the operation has no way of knowing.
Put the two records side by side, the way a plaintiff's firm does.
The first is the punch data, which the wage order compels: Wage Order 16 §6(A)(1) requires time records showing when each work period begins and ends, and adds that "meal periods, split shift intervals, and total daily hours worked shall also be recorded." Run the year and count the shifts over five hours where the meal started after the end of the fifth hour, ran short of thirty minutes, or does not appear. Under Donohue v. AMN Services (Cal. 2021), each of those is a record that raises a rebuttable presumption that the period was not provided.
The second is the premium ledger, which shows zero.
Read together, the employer's own records say: violations presumed on every one of those days, and not one addressed. That is the two-part structure we keep returning to, knowledge plus an absent documented response, assembled from the employer's own systems without a single deposition. And the premium does not travel alone: an hour owed and never paid is also a figure that never reached the wage statement, which opens a second question on the same day.
None of that requires bad faith. It requires an auto-deduct, or an unowned report, or an approval step, or a missing mapping, each of which was installed by someone trying to make payroll run cleanly.
Two cases, so the argument is not overstated.
Zero because the punches are clean. An operation where meals start before the fifth hour, run the full thirty minutes, and are punched, has nothing for the code to fire on. The punch data decides. A clean punch record beside an unused code is an unused code, and the test below will show it in a morning.
Zero because the exception was prevented. This is the outcome the whole loop is built for. An alert lands while the shift is still open, before the fifth-hour window closes, and the break is taken. The premium never becomes owed, and the code stays quiet for the right reason. Premiums are a cost to prevent, not a service to celebrate, and a code that fires often is not the target either.
The two zeros are told apart by one thing: the exception log. A prevented exception leaves an entry, the alert, who answered it, what happened. An unseen exception leaves nothing. An operation whose premium code and whose exception log are both empty has not prevented anything. It has not looked.
Every one of the five failure points has a structural answer, and none of them is expensive.
No auto-deduct: the meal is punched, or it is an exception. A clock prompt before the fifth hour, fired on unrounded punch times. An exception report every pay period with a named owner and a retained copy. A premium paid for every flagged shift, under its own code, tied to the specific date, so the ledger and the punches reconcile. A reporting channel for missed rest periods with a log, because no punch will ever exist for them, and a per-pay-period attestation that covers rest breaks, which is the only contemporaneous record that surface can have. No court has ever held that an employee attestation rebuts the Donohue presumption; what the attestation supplies is a dated, signed account where there would otherwise be silence. And corrections entered beside the punch they correct, never over it, so an approved timecard still shows what it said before approval.
That is the loop the rest-period post describes: the attestation captures it, a flag opens an item, the item carries a mandatory correction field, and nothing closes until someone records what was done. The premium code fires when it should, and the log shows why it did not when it should not.
Take the last full pay period and produce three counts.
If the first number is large and the second is zero, that is the finding, and it is a morning's work with two exports. If the first number is zero, the code is quiet for a reason the punches can show, and the exercise is finished. If the third number is zero in either case, nothing in the operation is recording a response, and that is the finding underneath the finding.
We read the three numbers with you on a free 30-minute call, and we say what the ledger shows next to the punches. Whether any particular shift was a violation, and what any hour of premium is worth, is your counsel's call.
Free · 30 minutes. If it fits, the next step is the records diagnostic, a fixed fee quoted on the call, and anything paid after that is a separate agreement you sign first.