Key takeaways
- Validate cleaning with residue data, not with visual inspection alone
- Use the hardest-to-clean product as the worst case, not the most frequent
- Swab corners and seams; rinse tanks and pipework for coverage
- Report swab results corrected for a measured recovery rate
- Revalidate after changing cleaning agent, product family or equipment
Cleaning between products is the control most likely to be described in a procedure and least likely to be proven by data. Validation turns the procedure into evidence: a defined worst case, a sampling method with known recovery, and a limit that is derived rather than copied from another site.
1. Why cleaning is validated, not inspected
A visually clean stainless surface can carry enough residue to affect the next batch’s colour, fragrance or preservative performance. Visual inspection only reaches accessible surfaces and only detects what is visible. Validation addresses the residue you cannot see, on the surfaces the product actually touches.
2. Choosing the worst case
Pick the product that is hardest to remove, not the one made most often. That is usually the most viscous, the most pigmented, the most fragranced or the one with the highest actives. If the worst case cleans to the limit, milder products are covered by the same procedure, which is why one validated case can replace dozens of individual studies.
3. Swab versus rinse sampling
Swabbing recovers residue from a defined area and is the better tool for corners, seams and mixer shafts. Rinse sampling covers large surfaces and hard-to-reach internals but dilutes the residue and gives no location information. Most programmes use both: swab the difficult areas, rinse the tank and pipework.
4. Setting a limit you can defend
Limits are usually derived from a health-based assessment of the previous product, or from a fixed fraction of a therapeutic dose where one exists. For cosmetics without such data, sites commonly use a limit tied to visual detection with a safety margin. The derivation matters more than the number: an auditor will ask how it was calculated.
5. Recovery studies
A swab result means nothing without a recovery figure. Spike a known quantity of residue onto the same surface material, let it dry, then swab and assay. Recovery of 70 to 90 percent is typical for stainless; porous or scratched surfaces recover far less. Report results corrected for recovery, not raw.
6. The areas that always fail
Mixer shafts, valve seats, O-rings, dead ends, the underside of baffles and the discharge chute account for most failures. These are also the places a swab can reach and a rinse cannot. A validation protocol that samples only the tank wall is testing the easiest surface in the plant.
7. Changeover between colours
Colour is the most visible cross-contamination marker and the cheapest to test. A spectrophotometer reading on a final rinse or on a first-off sample can confirm changeover without a full chemical assay. Many sites use colour as a rapid release test and hold the chemical results as the formal record.
8. Documentation
The record should carry the protocol, the worst-case rationale, the sampling map, the recovery study, the results, the deviations and the approval. A cleaning log with a signature proves someone cleaned. A validation report proves the cleaning works, and the two are not interchangeable.
9. Revalidation triggers
Revalidate after a change of cleaning agent, a new product family, a change in equipment geometry, or a run of failed routine checks. Annual review of routine results is normal. If routine swabs pass comfortably for a year, the interval or the agent can often be optimised, which is a cost saving rather than a risk.
10. Where it fits
Cleaning control matters most where products share a line, such as creams and lotions. Read more quality notes or see the range.
Reference: Quality systems notes, 2025
