Workshop hygiene and good manufacturing practice for small balm makers
Why an anhydrous balm still needs hygiene, how to sanitise equipment properly, and what ISO 22716 good manufacturing practice means in a home kitchen.
Balm making looks clean. No water, nothing curdles, nothing smells sour, and the whole process happens above 70 C. That persuades many makers hygiene here is a formality. It is not. The absence of water controls one hazard well and does nothing about another, and that difference is worth understanding before you write a cleaning procedure.
Low water activity suspends microbial growth, it does not sterilise. Spores and dormant organisms survive in fat indefinitely, and dry botanicals routinely carry them. So a balm workshop controls the starting load and keeps water out: clean, then sanitise with 70 percent alcohol, dry everything, separate from food and pets, keep batch records.
Why an anhydrous product still needs hygiene
An anhydrous formulation is genuinely inhospitable. ISO 29621 lists water activity below 0.75 as one of six factors qualifying a cosmetic as microbiologically low risk, and a balm with no aqueous phase qualifies. That is why balms do not need preservatives. Note what it does not buy you: qualifying removes the challenge test to ISO 11930, not the enumeration test to ISO 17516, because water activity does not correlate with microbial load. A dry product can carry a high count, and that count came from your workshop.
The mistake is reading "will not grow" as "is not there". FDA is blunt: below the minimum water activity for growth, bacteria do not necessarily die, and may remain dormant but infectious. Dry cosmetic raw materials frequently contain bacterial spores, and botanicals in particular cannot be assumed free of high counts of aerobic spore formers. Nor do spores merely survive: they germinate down to about 0.80 water activity, with Bacillus and Clostridium in the 0.89 to 0.94 range.
| Organism group | Minimum aw |
|---|---|
| Most spoilage bacteria | 0.90-0.91 |
| Staphylococcus aureus, growth | 0.83 |
| S. aureus, toxin formation | 0.85-0.88 |
| Salmonella | 0.94 |
| Clostridium botulinum, proteolytic types | 0.935 |
| Most moulds | 0.80 |
| Xerophilic moulds | 0.65 |
| Absolute observed floor for cell division | 0.605-0.64 |
The tahini experiment, and why water is the switch
The canonical demonstration is the 2008 to 2009 peanut butter outbreak. Salmonella persisted in a high-fat, low water activity food, survived 90 C for 50 minutes, and produced a final tally of 714 cases in 46 states, 166 hospitalisations and 9 deaths. Nothing in that product supported growth. It did not need to.
The cleanest analogue for a balm is not a cosmetic study either. Researchers followed Salmonella in tahini-based sauces of two kinds. In the water-based sauces, at a water activity of 0.997, the population rose by six logs in 24 hours at 22 C. In the oil-based sauces levels stayed flat. Same organism, same product family, same room. The fat killed nothing. It simply refused to switch anything on.
That is the situation inside a balm tin, and it explains why the hygiene priorities here are unusual. You are not preventing growth during manufacture, because growth will not happen in a hot fat melt. You are keeping the resting population low, because it persists, and because any water introduced later, from wet fingers, condensation or a shower shelf, is the switch.
The garlic-in-oil precedent makes the same point from the other direction. In a controlled study of chopped garlic in soybean oil (Solomon and Kautter, Journal of Food Protection, volume 51, 1988, page 862), one spore per gram produced toxin in 15 days at 35 C, and five spores per gram produced toxin at room temperature over 75 days, and even when highly toxic the garlic looked and smelled acceptable. The oil supplied the anaerobic conditions; the garlic supplied the water, the near-neutral pH and the spores. This is why herbal infused oils must be made with fully dried material.
ISO 22716 and what GMP means at small scale
ISO 22716 is the cosmetic good manufacturing practice standard, and the reference EU and UK law points to when it requires manufacture in accordance with GMP. It is not a laboratory specification but a set of expectations about personnel, premises, equipment, raw materials, production, quality control, waste, deviations, complaints, recalls and documentation. Almost all of it can be met in a domestic kitchen by a maker who writes things down.
The small-maker translation: know what you used, know what you did, know where it went, and be able to show all three a year later. Everything below implements that sentence.
Cleaning, sanitising and sterilising are three different things
These words are used interchangeably online and mean different operations. Cleaning physically removes soil, wax and oil. Sanitising reduces viable organisms on an already clean surface to a low level. Sterilising destroys all of them including spores, and needs heat or a sporicidal chemical, not alcohol.
The order is not optional. A sanitiser on a greasy surface is wasted, because alcohol will not penetrate a wax film. Clean first, and expect that to be the step that takes the time.
| Agent | Use | Contact time | Spores | Cautions |
|---|---|---|---|---|
| Hot water and detergent | Removing wax, oil and soil | Mechanical | No | Mandatory first step. Water hot enough to melt wax, above 70 C |
| Dishwasher, hot cycle | Tools, jugs, glass, silicone, stainless | Full cycle | No | Cleans and thermally disinfects in one pass |
| 70% ethanol or isopropanol | Sanitising clean dry surfaces and tools | 30-60 s wet | No | Flammable, keep off the hotplate. No rinse needed |
| Sodium hypochlorite, 200-1000 ppm | Sinks, drains, benches, spills | 1-5 min | At the higher end | Corrodes stainless and tinplate. Rinse and dry. Never mix with acid |
| Dry heat oven, 160 C for 2 h | Sterilising glass and stainless tools | 2 h at temperature | Yes | Heatproof items only. Not coated tins, plastics or seals |
| Boiling water, 10 min | Small tools where no oven is available | 10 min | No | Leaves items wet. Dry completely afterwards |
Why 70 percent alcohol beats 99 percent
This is the most common workshop error. Alcohol works by denaturing proteins, and denaturation needs water. Near 99 percent, the alcohol dehydrates the cell surface and coagulates the outer proteins into a shell that shields the interior, so the organism survives, and it evaporates almost instantly, removing the contact time the process needs. The effective band is roughly 60 to 80 percent, and 70 percent is the working figure.
Two limits. Alcohol is not sporicidal, which matters given that spores are what your dry raw materials most likely carry. And it only works on a visibly wet surface for the full contact time, so a wipe that dries in five seconds has done very little.
Decant 70 percent alcohol into a trigger spray and spray to visible wetness rather than wiping from a damp cloth. A cloth spreads soil and dries too fast. Let the surface air dry: buffing removes the contact time you were paying for.
Dry everything completely, because introduced water is the real risk
Given everything above, the single most important physical control in a balm workshop is that no water reaches the product. A drop of condensation in a lip balm tin creates exactly the localised high water activity pocket the whole anhydrous strategy exists to avoid, and it is the usual origin of mould in a balm.
Ambient humidity counts too, which is less obvious. Oils are mildly hygroscopic and drift toward the relative humidity around them. Measured on peanut oil, 722 ppm of water gives a water activity of 0.75 and 923 ppm gives 0.93, and oils carry 300 to 1000 ppm as supplied, so under a tenth of one percent water reaches the ISO line. Read that as an argument for a dry store and sealed drums, not as proof a balm is at risk: no published water activity values exist for shea, cocoa butter, beeswax or candelilla, and a lipid continuum offers no bulk water for a cell to occupy.
- Dry tools fully before use. Oven dry, air dry overnight, or alcohol dry: alcohol displaces water and evaporates cleanly, which is why it doubles as the final step.
- Keep water bath lid discipline. Steam condensing under a lid and dripping back into the melt is the commonest route for water into an anhydrous batch.
- Never wash tins or tubes. Buy them clean, store them sealed and inverted, sanitise with alcohol if you must. Washed containers are near impossible to dry inside the neck.
- Keep containers away from the sink. Fill on a bench with no splash path from a tap.
- Cool filled containers uncovered. Capping a warm tin traps vapour that condenses on the lid. Cool first, cap second, as covered under filling lip balm tubes.
Personal hygiene, hair and gloves
People are the largest single source of contamination in a small workshop, and S. aureus, carried by a substantial share of healthy adults, is the one common organism growing below 0.90 water activity. Wash hands with soap and warm water before starting and after any break, then glove up: gloves are additional to handwashing, not a substitute. Nitrile beats latex for allergy reasons, changed when torn, when they touch a face or phone, and between cleaning and filling.
Tie hair back and cover it. Remove rings and watches, which trap soil and water and cannot be sanitised on the hand. Wear a clean apron kept for making. Cover cuts with a waterproof dressing and a glove, and do not make product while ill. None of this is theatre: it is the cheapest available reduction in the starting count.
The workspace: separation, storage and pests
A domestic kitchen can be a compliant workspace if it is separated from food and animals in space or in time, and the separation is written down rather than assumed.
Pests deserve a mention because dried botanicals are the classic entry route. Pantry moths and beetles arrive inside dried herbs, not through the window. Freeze incoming botanicals for 48 to 72 hours to kill eggs, then store in sealed glass or food-grade plastic rather than paper. Inspect monthly, keep no open packets, keep dry goods away from the bench. The incoming goods routine is in sourcing ingredients.
Controlling water ingress after the product leaves you
The position on products preserved by low water activity is not "add a preservative". It is to control the packaging and the use. A German cosmetic expert working group states the rule plainly: products preserved with reduced water activity alone require waterproof primary packaging, especially if the product is likely to be used near water. FDA frames the consumer side: adding water or saliva to a cosmetic, and finger-dipping, are both listed contamination routes.
Those are packaging decisions, not formula decisions. Prefer a twist-up tube or a tight-lidded tin over an open jar for anything living in a bathroom, choose a closure that seals rather than covers, and say on the label to apply with clean dry hands or a spatula. Container trade-offs are in the packaging guide.
Adding an oil-soluble preservative as insurance is close to self-defeating. It partitions into the oil and will not reach water introduced later. Covering genuine water ingress needs a water-soluble preservative and a way to disperse it, a different formulation problem. That is reasoning from practitioners, not a written rule. Separately, EU Regulation 1223/2009 defines a preservative at Article 2(1)(l) and permits only Annex V substances as one. Tocopherol is not on Annex V.
Batch records, traceability and retained samples
Documentation is the part of GMP a small maker can achieve completely, and the part that matters if there is a complaint. A record has to answer two questions a year later: what went into this tin, and which other tins share its history. A batch code on every unit plus one record per batch does both.
| Field | Example entry | Why it is there |
|---|---|---|
| Batch code | 260906-LBV2-01 | Date, formula, sequence. Printed on every unit |
| Formula code and version | LBV2, rev 3 | Ties the batch to a specific approved formula |
| Raw materials | Beeswax lot BW-2411, shea lot SH-2503 | Traceability upstream if a supplier issues a notice |
| Actual weights | Beeswax 150.4 g against 150.0 g target | Records what you made, not what you intended |
| Process temperatures | Melt 78 C, 25 min, poured 68 C | Explains texture faults later |
| Yield and containers | 1000 g batch, 195 tubes, 12.1% loss | Verifies overage, flags a spill or short fill |
| Best-before assigned | 2027-09, from the shortest-lived oil | Shows the date was reasoned, not guessed |
| Retained samples | 2 tubes ambient, 1 frozen, shelf B | Physical evidence to re-examine after a complaint |
| Operator and deviations | Initials; "melt reached 88 C, batch scrapped" | Accountability and an honest failure log |
Retain at least two units per batch, one at ambient in the dark in final packaging and one frozen as a sensory control, until the assigned best-before has passed. Those samples are the backbone of shelf life and stability testing, and the only way to answer a complaint with evidence rather than apology. Keep the paperwork as long as your market requires, which in the EU and UK means the product information file obligations under selling balms in the UK and EU.
What a workable domestic routine looks like
Pulled together, a session runs: clear and clean the bench, clean and dry all equipment, sanitise bench and tools with 70 percent alcohol and let it evaporate, exclude pets and close the door, wash hands and glove up, weigh and record, melt and pour, cool uncovered, cap, code and label, then clean down and complete the record before leaving the room. It costs perhaps twenty minutes over working carelessly. The equipment list is on the equipment page, and the label content it feeds under labelling cosmetics in the US.
Start with the record. Of everything here, the batch sheet costs nothing, takes five minutes and cannot be reconstructed later. Write one for the next batch, even a hobby batch, and keep the retained sample beside it.
Frequently asked questions
If a balm has no water, why does hygiene matter?
Because low water activity suspends microbial growth, it does not sterilise. FDA puts it plainly: below the minimum for growth, bacteria do not necessarily die, and may remain dormant but infectious. Dry botanical raw materials frequently carry spores, spores germinate down to around 0.80 water activity, and any water that gets in later switches the population back on. Hygiene controls the load you start with.
How should I sanitise balm-making equipment?
In two stages. Clean first with hot water and detergent to remove wax and oil, because no sanitiser works on a dirty surface. Then sanitise the clean, dry surface with 70 percent ethanol or isopropanol, kept visibly wet for at least 30 seconds and allowed to evaporate. Dry everything completely before it touches product.
Why is 70 percent alcohol better than 99 percent?
Alcohol kills by denaturing proteins, and that reaction needs water. Concentrations near 99 percent dehydrate the cell surface, coagulate the outer proteins into a shell that protects the interior, and evaporate too fast to give a useful contact time. The 60 to 80 percent band, with 70 percent as the working figure, is both more effective and slower to dry.
Do I need ISO 22716 certification to sell balms?
Not certification, but you do need to work to the principles. ISO 22716 is the cosmetic good manufacturing practice standard, and EU and UK law requires manufacture in accordance with GMP, with ISO 22716 as the recognised reference. A small maker demonstrates that through documented procedures, batch records, traceable raw materials and retained samples rather than through an audit certificate.
Can I legally make cosmetics in a home kitchen?
In most jurisdictions yes, provided the space is suitable and the practice is defensible. What matters is separation from food preparation and pets in time or space, cleanable surfaces, dedicated equipment, controlled storage and written records. A kitchen used for cosmetics between meals with dedicated tools and a documented clean-down routine is workable. A cluttered shared bench is not.
Does dried plant material in infused oil need extra controls?
Yes. Drying removes the water that organisms need, but it does not remove the spores. Use fully dried material only, dry it to the point of brittleness rather than leathery, freeze incoming botanicals for 48 to 72 hours to kill insect eggs, and store them sealed and dark. CDC advises refrigerating homemade garlic or herb oils and discarding them after four days.
What should a batch record contain?
A unique batch code, the date, the formula code and version, every raw material with supplier and lot number, the actual weights taken rather than the targets, melt and pour temperatures, the containers filled and the count, the yield achieved, the best-before assigned, where the retained sample is stored, who made it, and any deviation with what you did about it.
Sources and further reading
- International Organization for Standardization, ISO 22716:2007, Cosmetics: Good Manufacturing Practices (GMP), Guidelines on Good Manufacturing Practices.
- International Organization for Standardization, ISO 29621:2017, Cosmetics, microbiology, guidelines for the risk assessment and identification of microbiologically low-risk products, and ISO 17516 on microbiological limits.
- European Union, Regulation (EC) No 1223/2009 on cosmetic products, Article 8 on good manufacturing practice and Annex V on permitted preservatives.
- US Food and Drug Administration, Hazard Analysis and Risk-Based Preventive Controls for Human Food: Draft Guidance for Industry, Appendix 3, Table 3-A, on minimum water activity for pathogen growth.
- Centers for Disease Control and Prevention, Guidance on home-prepared garlic and herb-infused oils, on refrigeration and four-day discard.
- Survival and growth of Salmonella in oil-based and water-based tahini sauces, Microbiology Spectrum, 2026.
- US Centers for Disease Control and Prevention, Multistate outbreak of Salmonella Typhimurium infections linked to peanut butter, 2008-2009, final case counts.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.