A retain sample is the cheapest insurance a factory can hold. Kept properly, it settles a complaint in a day. Kept badly, it proves nothing and can make a defensible position look careless. Quantity, storage and duration are the three decisions that matter.
1. Two sample types, two purposes
A control sample represents the lot as shipped and stays for reference. A stability sample is pulled at intervals and tested against a protocol. Mixing the two in one cupboard is common and unhelpful, because one must stay untouched while the other gets opened. Label them separately and store them apart.
2. How many units
Two to three retail units per lot is a common minimum, plus one bulk sample of 100 to 200 grams where the product is filled from bulk. Keep one unit sealed for the full period and allow the second to be opened for inspection. For lots under 500 units, one retail unit plus bulk is usually accepted if the reasoning is written down.
3. Where they live
Twenty degrees plus or minus five, out of light, out of the shipping warehouse, in a room with a logged temperature. A shelf above a filling line is not storage. Excursions should be recorded rather than hidden, because a documented short excursion is defensible and an unlogged one is not. Keep the log with the samples, not in an office.
4. How long to keep them
A common rule is shelf life plus twelve months counted from the fill date, and some export customers ask for shelf life plus two years. Where the product carries a period-after-opening figure, keep at least one unit beyond the longest market’s shelf life. Where a customer audits the plant, agree the number in the quality agreement rather than leaving it to habit. Review the retention period annually against the markets actually served.
5. What a sample can prove
It can confirm identity, appearance, odour, label version and fill weight at the time of manufacture. It can be re-assayed for an active if the analytical method still works on aged material. It can be photographed beside a returned unit to show whether a complaint describes a change or a product that always looked that way.
6. What it cannot prove
A retain sample cannot show whether the batch was microbiologically clean at fill, because storage changes the population in both directions. Nor can it show that the process ran to specification, because it holds no process data. These gaps are why the batch record and environmental monitoring matter as much as the sample itself. A sample without its batch record answers the wrong question.
7. Seals and chain of custody
Seal each unit with a tamper strip carrying the lot number, and record who sealed it and when. When a sample is opened, note the date, the reason and the person. This is a five-second habit that turns a jar on a shelf into evidence. Without it, the other side can reasonably question whether the sample is the one that shipped.
8. When a complaint lands
Retrieve the sample within 24 to 48 hours, photograph it under consistent lighting beside the returned unit, and record weight and appearance before opening. Do not break the seal until comparison photos exist. Most disputes settle at this step, before any analytical work, because the two units usually either match or clearly do not.
9. What the programme costs
Storage for a mid-sized factory is a few square metres and a logging thermometer. Against the cost of a recall, a write-off or a lost account, the programme pays for itself the first time it is used. The failure mode to avoid is a large sample store with no index, where nobody can find the lot that matters within a day.
10. Where it fits
Retention practice supports release testing across creams and serums alike. Related quality notes are in encyclopedia.