Handling a shelf life complaint and working out whether your date was wrong

Most premature failures come from a date copied off a supplier sheet. Peroxide bands, the shortest oil rule, retained samples and how to redate honestly.

A customer writes to say the balm smells of crayons and it is only nine months old, and the tin says eighteen. Two quite different things can have happened, and they need opposite responses: either that particular tin was stored somewhere hot and bright, or the date on every tin in the batch is wrong. This page is the procedure for telling those apart, deciding what to do with the stock you still hold, and setting a date you can defend next time.

• PV under 10 meq/kg
• Shortest oil sets the ceiling
• Retain two units per batch
• Antioxidant 0.05 to 0.2%

What the complaint is actually reporting

Start by working out what the customer smelled or saw, because several unrelated faults arrive worded as "it has gone off". A balm has no water in it, so it does not spoil in the way food spoils, and most reports of spoilage in an anhydrous product are one of four things.

Only the first row is a shelf life failure. The rest are faults that a longer date would not have prevented and a shorter date would not have fixed. Ask for a photograph and ask where the tin has lived, and ask both before you offer anything, because the answer changes what you owe the customer and what you have to do about the remaining stock.

Open the retain, which is exactly what it is for

The single most informative thing you can do costs you one unit. Take the sealed retained sample from the same batch, which has sat in the dark at a steady temperature since the day you poured it, and open it alongside a frozen control of the same formula. Warm both to room temperature first, then smell each on the back of the hand after about ten seconds, look at the colour against white paper, and press the surface.

If you have no retain, you have no diagnosis, only opinions. Two units per batch in final packaging, unopened, dark, at a steady 15 to 20 C, kept until the best before date plus six months, is the standard set out under batch records and traceability. The six month tail exists precisely for this situation, because complaints about a date arrive after the date has passed.

Peroxide value, and the bands worth acting on

If the complaint has commercial weight, or you need to know how far along the oil phase is, send a sample for a peroxide value. It is measured by iodometric titration under ISO 3960 or AOCS Cd 8b-90 and reported in milliequivalents of peroxide oxygen per kilogram of oil. Codex sets legal ceilings for edible oils at 10 meq/kg refined and 15 meq/kg for cold pressed and virgin grades, and those ceilings are the origin of the working bands used in trade.

The chemistry behind those numbers, and why the count of bis-allylic hydrogens predicts the rate far better than the count of double bonds, is set out under rancidity and oxidation. The practical point here is narrower: a laboratory number confirms or refutes a nose, it does not replace one, and it costs more than a batch of hand balm is worth unless there is a real dispute.

Where your date came from, and why twelve months is not evidence

Most premature failures are not chemistry going wrong. They are a date that was never derived from anything. Three habits produce dates that fail early.

1. Copying the longest figure on the supplier sheet. Vendor shelf life figures for oils carry no stated method, storage temperature, packaging or endpoint. A sheet saying "24 months" is a hope about a sealed drum in a cool warehouse, not a prediction about your tin on a customer's shelf.
2. Averaging the oils. A blend does not last as long as its mean component. The least stable oil generates the free radicals that then attack everything else, so a blend with 10 percent rosehip in it is a rosehip-limited blend. Run the fatty acid arithmetic through the oil blend calculator and look at the linoleic and linolenic totals, not the average of the vendors' claims.
3. Defaulting to twelve months because everyone does. Twelve months is a convention, not a measurement. It is generous for a blend heavy in polyunsaturates and needlessly short for one built on jojoba, squalane or high oleic oils. Either way it tells the customer nothing you have checked.

The date also has to start from the right point. Oil that was already eight months old when it arrived has spent eight months of its life, and your batch inherits that. Ask for the peroxide value and the date of analysis at the point of purchase, which is the part of sourcing ingredients that pays for itself. Then subtract the stock age from the ceiling, apply a margin, and claim less than you believe. The shelf life calculator does that arithmetic, including the subtraction people forget.

Storage abuse: the three places balm actually dies

When the retain is clean and the customer's unit is not, the cause is almost always one of three locations. None of them are unusual, which is why a date has to carry a margin for them rather than assume they do not happen.

Two of those involve light as well as heat, and light matters more than most makers expect. Real time storage work on unstable oils has found the container and the light regime separating a sample that stayed under 10 meq/kg for nine months from an identical oil that passed 40 meq/kg in five, with no difference in fatty acid profile at all. That is why packaging choice is a shelf life lever and not a cosmetic one, and why the advice in storing balms at home belongs on your insert card rather than buried on a website.

A repeating pattern of complaints from one region in one season is a different signal again. That is usually a distribution problem, and the fix is in dispatch scheduling and packing rather than in the formula. Shipping in hot weather covers the route side, including what a summer courier network does to a pallet of stock.

The 40 C oven test, and what it can honestly tell you

The common small-maker protocol is twelve weeks at 40 C in the dark as a rough proxy for a year at ambient. The arithmetic behind it is a Q10 assumption: reaction rates roughly double for every 10 C rise, so 40 C against a 20 C ambient is two doublings, a factor of four, and twelve weeks multiplied by four is about eleven months. That is where the proxy comes from, and stating it out loud is the point, because the assumption is doing all the work.

It breaks in three specific ways for a balm, and none of them are minor.

• The physical state changes. A balm that is solid at 20 C may be soft or fully liquid at 40 C. Oxygen diffuses through a liquid far faster than through a crystalline wax network, so the 40 C sample is not the same system running faster, it is a different system. Batches with low solid fat content are affected most.
• Antioxidants deplete faster than the substrate oxidises. Tocopherol consumption is not simply proportional to the oil's oxidation rate, so the accelerated sample can lose its protection early and then fail out of proportion to real time.
• The dominant mechanism shifts with temperature. This is the reason induction periods measured at 100 to 130 C cannot be converted into months either, and the same objection applies, in milder form, at 40 C.

What the oven does reliably is expose the failures that arrive before oxidation does: separation, sweating, grain, colour drift, label adhesive lifting and packaging attack. Those are what fail most handmade batches in practice. The protocol for running it alongside real time samples is in shelf life and stability testing.

What to do with the stock you are still holding

Three outcomes, chosen by what the retain told you.

Relabelling is legitimate only downwards, and only to a date your own evidence supports. Shortening a date is a correction; extending one, or reprinting a batch with a later date because it still looks fine, is not something the evidence can support and not something to do.

Can you remelt it?

Only if oxidation has not started. This is the one question where the honest answer is narrower than makers want.

Rancidity is not reversible. The primary hydroperoxides have already fragmented into aldehydes and ketones, and heat accelerates exactly that decomposition, so a rescue melt on a turning batch produces smooth balm that smells worse than it did before you started. Nothing is filtered out, nothing evaporates in useful quantities at balm melt temperatures, and adding tocopherol at this stage does nothing, because an antioxidant intercepts propagating radicals and cannot repair the products of past propagation. Rancid balm covers the smell-by-smell identification and the safety question in full.

Rework is reasonable in exactly one case: the retain is clean, the peroxide value is low, and the fault is physical. A batch that grained, sweated, tunnelled or set too soft can be remelted, corrected and repoured, and it is not a compromised batch. If you do rework, the reworked material carries a new batch code linked back to the original, its shelf life runs from the original oils rather than from the rework date, and it gets its own retains. The link back to the original batch code is the part that gets forgotten, and it is the part that matters if the reworked stock is ever questioned.

Preventing the next one

Four levers, in the order they actually matter for a batch that failed early.

Take the antioxidant rates as a ceiling rather than a target. Tocopherol is well characterised, it is the antioxidant most balms should use, and it protects the oils in the tin rather than the skin, which is a distinction worth being clear about on a label. Both the mechanism and the pro-oxidant reversal at high loading are covered under vitamin E and antioxidants.

One more prevention that is not chemical: put the storage advice where the customer will read it. Most of the abuse cases in the table above are avoidable by a line on the base label. Keep out of direct sunlight, do not leave in a vehicle, use within six months of opening.

The limits of what you can conclude

Two honest boundaries. The first is that a shelf life complaint from a single customer, with no retain and no batch code, is not diagnosable. You can be courteous and replace the product, and you cannot know whether the date was wrong. That is an argument for the record keeping, not for a longer investigation.

The second is that the date on a cosmetic is a claim you are responsible for, and no test available to a small maker converts cleanly into it. Under EU and UK rules a product with a minimum durability of 30 months or less must carry a date of minimum durability, and most handmade balm falls under that line, so the date is not optional. The obligations that come with selling are set out under selling balms in the UK and EU. The decision rule that follows is uncomfortable and correct: claim the shortest date your evidence supports rather than the longest date you can imagine defending, and shorten it the first time a retain tells you to. A conservative date costs you a little stock rotation. A generous one costs you the complaint you are reading now.

Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.