Chasing a soapy or bitter taste back to the ingredient that actually caused it

Free fatty acids, unsaponifiables, sweeteners and detergent residue all taste different. A tasting ladder that isolates the culprit and the swaps that clear it.

Soapy, waxy and bitter are three separate faults with three separate causes, and the fix for one does nothing for the others. Most makers reach straight for the flavour oil, which is almost never responsible. The fault is usually upstream of the formula altogether: an oil that has hydrolysed, a jug that was not rinsed, or a butter whose taste was always going to come through.

• Crude shea unsaponifiables: 5-11%
• Stevia window: 0.05-0.25%
• Flavour oil: 0.5-1%
• No remelt fixes taste

Soapy, waxy and bitter are not the same complaint

Getting the descriptor right halves the diagnostic work, because each one points at a different chemistry. Taste at skin temperature, on the inner lip rather than the tongue, and wait a full minute: most of these faults live in the finish rather than the first impression.

Soapy is a specific thing, not a synonym for unpleasant. It is the flavour of free fatty acids in the C8 to C12 range, sometimes with a slippery, filmy mouthfeel alongside it. If you have ever tasted a coconut fat that has been kept warm too long, that is the note exactly. A surfactant film produces something similar, which is why the two causes get confused.

Waxy is usually not a taste at all. It is a coating sensation with very little flavour attached, and it comes from a high wax or high stearin content sitting on the lip. People report it as a taste because the sensory system is bad at separating the two. If the complaint disappears the moment the film is wiped off, you are looking at texture, and the same material is probably causing the drag and poor glide that comes with it.

Bitter is a genuine taste, perceived at the back of the tongue and slow to clear. In a balm it comes from unsaponifiable matter, oxidation products, overdosed sweeteners or excess tocopherol. It also varies enormously between people, which matters when a single customer complains.

Free fatty acids, and why lauric oils go soapy specifically

Triglycerides hydrolyse. Given traces of water, warmth, time or a surviving lipase, the ester bonds split and release free fatty acids into the oil. This is lipolysis, and it is a different reaction from the oxidation covered under rancidity and oxidation, though a neglected oil usually suffers both.

What makes it a taste problem rather than a chemistry footnote is chain length. A free oleic acid molecule, at eighteen carbons, is barely detectable by mouth. Free caprylic (C8), capric (C10) and lauric (C12) acids are volatile enough and low enough in flavour threshold to be obvious at fractions of a percent, and their descriptor across the food science literature is exactly the one makers use: soapy. That is why coconut oil, fractionated coconut, palm kernel and babassu are the usual suspects, and why an olive or sunflower based balm that goes off tastes sharp and cardboard-like instead.

The measurement is the acid value: milligrams of potassium hydroxide needed to neutralise the free acids in one gram of fat, determined by titration under ISO 660. Free fatty acid percentage is the same information in different units, and for an oleic-type oil the conversion is roughly acid value times 0.5.

Two practical consequences. Free fatty acids are already present in the oil when you buy it, so an oil bought cheaply in a large drum and used slowly is the classic source, which is the trade-off examined under shelf life testing. And they keep forming in the finished balm, so a taste fault that was absent at pouring and obvious at three months is almost always this. Water is the accelerant, which is one more argument for keeping every trace of it out of an anhydrous product.

Detergent residue, the cause almost nobody checks

This is the single most commonly missed explanation for a soapy note, and it is easy to prove or exclude in an afternoon.

Anionic surfactants of the kind in washing-up liquid are detectable by mouth at very low concentrations, and they cling to exactly the materials a balm workshop uses: polypropylene jugs, silicone moulds and spatulas, wooden spoons, and the roughened inside of a well-used stainless pan. A film left after a quick rinse dries down and stays. Hot oil then does what water could not, dissolving and redistributing it through the batch. The result reads as soapy because it is soap.

Dishwashers are worse, not better. Rinse aid is a nonionic surfactant that is deliberately designed to remain on the surface as a thin film so that water sheets off rather than spotting. Doing its job correctly means leaving residue behind, and a balm jug run through a domestic dishwasher comes out coated on purpose. Silicone moulds are the worst offenders of all, because silicone is porous to small molecules and holds both surfactant and flavour from previous batches.

Unrefined shea and the unsaponifiable fraction

Crude shea butter carries an unusually large unsaponifiable fraction, commonly reported at 5 to 11% of the butter and varying widely with provenance and processing. Almost every other cosmetic fat sits below 2%. That fraction is mostly triterpene alcohols and their cinnamate esters, along with sterols, hydrocarbons and tocopherols, and it is where shea's smoky, nutty, distinctly bitter character lives. At 20% of a lip balm you are putting roughly 1 to 2% of that material directly on the lip, and it does not hide.

Refining changes this less than most people assume, and in a specific way. Bleaching removes colour and some polar impurities, and deodorisation strips volatile odour and flavour compounds by steam under vacuum at high temperature. What deodorisation does not do is remove the non-volatile triterpene alcohols, which largely survive. Refined shea therefore tastes close to neutral because the volatiles are gone, not because the unsaponifiables were removed. That distinction matters when someone sells you an unrefined butter as being richer in the good fraction: they are right about the chemistry and it is still the wrong choice for a lip product.

If bitterness is the complaint and shea is in the formula at any meaningful level, the fastest test is to swap to a refined grade or to a butter with a smaller unsaponifiable fraction, as compared in butters compared. Cocoa butter brings its own strong flavour but a pleasant one; kokum and mango are close to flavourless.

Beeswax, tallow and lanolin: what refining really changes

Read down the last column, because that is where the honest answer sits. Refining does most for shea and least for lanolin. Beeswax in a white filtered grade is close to flavourless and yellow beeswax is not, so a honeyed note in a lip balm is a grade decision rather than an accident. Tallow that was rendered wet, washed and kept cold is mild; tallow rendered hot with connective tissue still in it is not, and no downstream step in your workshop repairs that. Lanolin is the material where expectations most need managing: pharmaceutical grades are far cleaner than crude wool grease and still taste of themselves, so if a customer objects to the taste of lanolin, the answer is to remove it rather than to buy a better one.

Sweeteners have a narrower window than the marketing suggests

Stevia extract powder sweetens a balm at roughly 0.05 to 0.25% of the batch. The upper end of that window is not a safety limit, it is a taste cliff. Steviol glycosides carry a bitter, liquorice-like back-note that appears abruptly rather than gradually, and past about 0.25 to 0.3% it dominates whatever flavour you paid for. Because the dose is so small, a scale reading only to 0.1 g cannot hit the window at a 30 g test batch at all, and overshooting by a factor of three is easy. The predispersion trick and the full rate table are on lip balm flavour and sweeteners.

The other reliable mistake is trying to fix a bad base with more flavour. Flavour oil at 0.5 to 1% of the batch is a normal, effective level. Doubling it does not mask a soapy or bitter base, it stacks on top of it, and compounded flavour oils bring their own bitterness at high levels. If the base is wrong, correct the base.

The tasting ladder

Guessing costs more than testing. This takes about two hours across two sessions and it isolates the cause definitively.

1. Taste every raw material on its own. A smear of each fat, wax and oil on the inner lip, at room temperature, with plain water and a dry cracker between samples. Wait a minute after each. Write down a descriptor before you move on, because the memory for taste is short and unreliable.
2. Build a null base. Twenty grams of the two most neutral materials in your cupboard, typically a white filtered beeswax and a fractionated coconut or jojoba, melted in glass and poured with tools you have just wiped with isopropanol. Nothing else at all.
3. Taste the null base after 24 hours. If it tastes soapy, your problem is equipment or one of those two materials, and the formula is innocent.
4. Run the equipment control. Pour the same null base twice: once from your usual jug washed the usual way, once from a jug rinsed to no foam, dried and alcohol wiped. If they differ, you have found detergent residue and you can stop.
5. Add back one suspect at a time. Fresh 20 g null base for each, with a single formula ingredient included at its real percentage. Four to six mini-batches covers most formulas. The one that reproduces the fault is the cause.
6. Blind the final comparison. Have someone else code the samples and hand them to you in random order. Expectation drives taste perception hard enough to invalidate an unblinded test, which is the whole reason for the protocols in sensory testing.
7. Retaste the confirmed cause at three months. Hydrolysis faults grow. If a sample that was clean at pouring is soapy at twelve weeks, the mechanism is time, not contamination.

Reading the result

What can and cannot be rescued

Start with the hard part: no remelt cures a taste fault. Removing volatile off-flavours from a fat is deodorisation, and industrial deodorisation means steam sparging under high vacuum at 180 to 240 C, which you cannot reproduce and which would destroy the balm if you could. Reheating to 70 C in a water bath drives off nothing that matters. Neither does adding more flavour, more sweetener or more antioxidant.

Three faults do have a partial route out. An overdosed sweetener is dose dependent, so remelting the batch with an equal weight of unsweetened base halves the level, which rescues it if you overshot by roughly a factor of two and you have containers to fill. A texture fault mistaken for a taste fault is a straightforward reformulation, and it can be corrected in a remelt using the ratios in how to make lip balm. And a batch that only fails as a lip product can be repurposed as a cuticle or hand balm, where taste is irrelevant, provided you relabel it honestly and the change of use does not take it outside whatever safety assessment covers it.

Everything else is a loss. Detergent contamination is distributed through the fat and cannot be removed. Free fatty acids will not re-esterify. Oxidation products are permanent, and the batch belongs in the same category as any other rancid balm. Keep the failed batch for your own use if it is merely unpleasant, and discard it if it smells actively off, but do not sell it. A taste fault is the one defect a customer notices immediately and never forgives.

Preventing the next one

Most of this is bought rather than made. Refined grades for anything going on lips: refined shea, white filtered beeswax, refined or fractionated coconut. Buy lauric oils in quantities you will use inside six months, and store every oil cool, dark and closed, because headspace oxygen and warmth drive both spoilage reactions. Smell each oil on arrival and again before each batch, since the nose detects both hydrolysis and oxidation earlier than the mouth does.

Then the process. Dedicated lip-product equipment, washed with hot water only and finished with alcohol. No dishwasher, no rinse aid, no silicone moulds carried over from a scented batch. Flavour oil at 0.5 to 1%, stevia from a predispersion or not at all, tocopherol at 0.1 to 0.5% of the oil phase. Keep a retained sample from every batch and taste it at ten days and at three months, so that a slow hydrolysis fault reaches you before it reaches a customer. That is the same retain that shelf life testing already asks you to keep, so it costs nothing extra.

Where your own palate stops being evidence

Taste perception varies more between people than any other sense used in this work. Bitter sensitivity in particular is strongly genetic, and the same balm can be unremarkable to you and unpleasant to a fifth of the people who try it. A single complaint about bitterness is therefore weak evidence about the product and strong evidence about that person, while three complaints out of ten is a formulation problem regardless of what you taste.

Expectation matters just as much. If you know a sample contains the ingredient you suspect, you will taste it, which is why the blinded step in the ladder is not optional theatre. Two or three assessors, samples coded by someone else, and a written descriptor recorded before discussion, is the minimum that produces usable information.

The decision rule is short. If the fault appeared immediately and affects everything you make, look at your equipment first and your grades second. If it appeared over time, look at the oils. If it is bitter rather than soapy, look at unrefined shea and at the sweetener before anything else. And if you cannot reproduce the fault in a blinded test with a null base, do not reformulate: you are chasing a difference that may not be in the product.

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