A cleaning validation has to prove a number, not a look

211.67 sets the duty; it doesn't set the number

21 CFR 211.67 requires equipment and utensils to be cleaned, maintained, and sanitized at appropriate intervals to prevent contamination that would alter a drug product's safety, identity, strength, quality, or purity, and requires written procedures for how that cleaning gets done, including the removal or inactivation of residues left by whatever ran through the equipment before. The regulation states the duty in general terms and leaves the acceptance criteria — how clean is actually clean enough — to the manufacturer to establish and defend, which is exactly the gap a cleaning validation protocol exists to close.

The 1993 guide set the expectation that cleaning gets validated at all

FDA's Guide to Inspections of Validation of Cleaning Processes, issued in July 1993, is written for the agency's own investigators rather than industry, and says so on its face — it doesn't bind FDA or create rights for anyone reading it. What it does is state plainly that a firm is expected to have validated its cleaning process, not merely performed it consistently, and that the validation has to rest on scientific data showing the process reliably meets predetermined specifications rather than on a visual check alone. More than three decades old, it remains a baseline inspectional expectation this area still gets measured against.

Visual and analytical are two different tests, not two options

A visibly clean surface is a real, necessary checkpoint — residue above a certain level is usually visible to a trained eye, so a visual-inspection failure is an immediate, unambiguous fail. It's also not sufficient on its own, because a residue level well below what's visible can still exceed a limit that matters: a potent compound's safe carryover threshold, or a sensitizing material's, can sit far under the concentration a surface would ever show. A cleaning validation protocol has to pair the visual check with an analytical one — commonly a swab or rinse sample tested for a specific residue marker — precisely because the two methods catch different failure modes at different concentrations.

The limit is the manufacturer's to set and defend, not a fixed FDA number

ICH Q7's own section on cleaning validation, written for active pharmaceutical ingredient manufacturers, asks for residue limits that are practical, achievable, and verifiable, and based on the most harmful residue likely to be present — without naming a fixed carryover number of its own. The commonly cited rules of thumb in this space, a maximum of 10 parts per million or one-thousandth of a therapeutic dose carried into the next batch, come from industry practice and guidance documents outside ICH Q7 itself, not from Q7's own text. A validation protocol that cites a specific numeric limit has to show where that number came from and why it's conservative enough for the specific residue and the specific next product — not treat 10 ppm as a figure ICH itself requires.

Where this meets the file

A cleaning validation protocol needs the acceptance criteria — the residue identified, the limit set, and the rationale behind that limit — recorded as its own section, separate from the sampling method and the visual-inspection result, so an inspector can trace the number back to the reasoning rather than finding a limit that simply appears. A protocol template built around that structure, alongside the validation master plan that schedules it, is previewed in the launch catalog. If your site sets these limits differently, the shelf takes that correction directly.