California's meal-period argument has run through the per-pay-period record since 2021. Four years after Donohue v. AMN Services, what that record looks like in operating terms — and what no court has yet held about it.
The California Supreme Court decided Donohue v. AMN Services in February 2021. The holding has been settled law for four years. The operating implications are a different matter, and they are still the part that catches operators out.
The holding, in plain language: time records showing missed, short, or delayed meal periods create a rebuttable presumption that the meal period was not provided. The records the Court was looking at were rounded — clock-ins and clock-outs aggregated to the nearest ten-minute increment — and the Court held that employers may not round time punches in the meal-period context at all, because the rounding operates on a right that is measured in minutes.
The presumption is rebuttable. That word is doing the work in every defense built on Donohue since.
A presumption is a legal default. It tells the fact-finder which party loses if no further evidence is produced. With the rebuttable presumption Donohue established, the burden falls to the employer to come forward with evidence that the meal period was either taken, knowingly waived, or excused under one of the recognized exceptions. The employer doesn't have to prove the negative absolutely. It has to produce enough evidence to shift the burden back to the employee to prove the violation.
The form of evidence defense-side practice has consolidated around since is contemporaneous per-pay-period employee attestation. Each pay period, the employee affirms or disputes specific facts about meal and rest breaks taken during that period. The attestation is dated, signed, and stored.
The pedigree of that idea is worth stating precisely, because it is routinely overstated. The employer in Donohue ran a per-period electronic system that prompted employees about their meal periods. The Court described it, and did not condemn it as a category. It failed there because the timekeeping underneath it rounded punches, so the prompt fired against numbers that had already erased the short break it was supposed to surface. The most that can honestly be claimed is this: the California Supreme Court described an employer system of the same per-period design and said that, without the rounding, it "would have ensured accurate tracking of meal period violations." We don't round.
What cannot be claimed is that attestation is the tool the Court prescribed. Donohue expressly left the sufficiency question for remand, and no court has ever held that an employee attestation rebuts the Donohue presumption. We disclose that on every version of this argument we make, because a vendor who does not disclose it is either unaware of it or hoping you are. An attestation is an argument with a good pedigree and no holding behind it. That is a real asset. It is not a defense, and the distinction matters most on the day it is tested.
A defensible per-period attestation, in operating terms, captures four things at a minimum:
Each of these is a specific factual claim about a specific shift on a specific day, attested by the specific employee whose break it was. The attestation does not need to be a single document. It needs to be a record system that can produce, on demand, for any employee on any pay period in the relevant statute-of-limitations window, the four facts above.
The most common defense artifact employers point to is a meal-period waiver signed at hire. The waiver says some version of I acknowledge the company's meal-break policy and agree to take my breaks as required.
This is not a Donohue defense. It is a policy-distribution document. It establishes that the employee was informed of the rule. It establishes nothing about whether any particular break was taken on any particular day. The presumption is per-period; a one-time hire-day waiver is not.
Plaintiffs' counsel know this distinction. So do the courts. The presence of a hire-day waiver does not defeat a Donohue presumption; the absence of it is a separate problem on a separate claim.
Three failure modes account for most of what goes wrong:
Batch attestation. The employer collects attestations once a month, or once a quarter, summarizing what happened in the period. Late-batched attestations are weaker than contemporaneous ones. By the time a quarterly review prompts an employee to sign, the employee no longer remembers which shift on which day featured which break. The attestation devolves into a check-box exercise that does not survive cross-examination.
Manager-signed attestations. The employer collects a single signoff from the foreman, the supervisor, or the office manager, claiming that breaks were taken across the period. This is a witness statement, not an attestation. The person with personal knowledge of whether a given break was taken, and whether it was duty-free for the full thirty minutes, is the employee who did or did not take it. A manager's signoff asks a witness to testify about someone else's afternoon.
Aggregate attestations. In Q2, all my breaks were taken on time. The employer wants efficiency; the employee wants to finish the form; the law wants per-shift specificity. Aggregate attestations cannot rebut a per-shift presumption because they cannot speak to whether the specific shift the plaintiff identifies as the violation date was the one that was anomalous.
The structural fix is per-period, per-employee attestation tied to specific shifts, captured contemporaneously, stored in a way that produces a clean record on demand. Bilingual where the workforce is bilingual. Recorded against the time data, not separately from it.
This work is what Floburn ships as MicroForensics for California construction and logistics employers. The platform is built on the timekeeping system the employer already runs — BusyBusy, ExakTime, ADP, Paylocity, Samsara, Motive, or whatever combination is in place, with the connector built per stack — and routes a per-period attestation flow to every hourly employee. Deployment includes a rounding check on the timekeeping underneath: attestation prompts fire on unrounded punch times — where the customer's system rounds punches, the rounding is turned off, and where unrounded punches cannot be supplied, we say so rather than prompt off rounded data. The output is a structured, queryable record that produces, on demand, the per-period evidence a Donohue rebuttal argument is built from — an argument with a good pedigree and no holding behind it.
It also runs the four modules §2699(g)(2) actually enumerates — the recurring payroll audit with action taken on findings, the policy dissemination and acknowledgment record, the supervisor training record, and the corrective-action log. Those are the reasonable-steps instruments. The attestation is not among them, and we keep the two straight deliberately: the attestation is a Donohue evidentiary argument, and the enumerated modules are what a §2699 reasonable-steps showing is built from. They are produced by one system and they do two different jobs.
What links them is not doctrine but architecture. Both are records with a correction attached — an exception surfaced, and a dated entry saying what was done about it. That architecture is what holds in every direction this analysis runs, and the correction is the half most employers are missing.
Whether a given attestation regime is sufficient is a legal conclusion about specific facts, and not one we reach — that question goes to your counsel, and so does the language on the form. What an operator can observe without any legal conclusion at all is whether the four facts above are recoverable, per employee, per period, on demand. If they are not recoverable, no argument gets built on them later, whatever a court eventually says the standard is.
The forward-going record can be built in days; the historical one cannot be rewritten. And the trigger is earlier than most operators plan for — a records request under section 226, 432, or 1198.5 closes the window well before a complaint does, and the records request usually comes first.
If you'd like to walk through how this looks against your current timekeeping stack, the discovery call is the right starting point. If your existing system already produces the per-period record, we will say so and we will not sell you a second one. We will tell you what we find.