My Fragrance Continues Separating Even at a Lower Load — When Should I Stop Trying to Rescue the Formulation and Replace the Fragrance Oil?

Lowering the load is step one, not the whole plan

When a candle sweats, cutting the load is the right first move. It is cheap, it is fast, and in a large share of cases it works: the oil was simply above what that wax holds. You should be suspicious of any advice, including a supplier's, that jumps straight to selling you a new oil. CSI sells oil, and the honest default is still to use less of it first.

But load is only one of the causes on the list. Oil composition, under-mixing, adding oil too cool, fast or uneven cooling, and storage heat can all push fragrance to the surface. If you keep lowering the load and the oil keeps appearing, each new round is a week or two of observation with no new information. At some point the formulation has told you what it can tell you, and the decision is whether the oil stays.

The test that makes the decision possible: a control oil

Most makers who get stuck have run their load ladder with only the problem oil. That leaves two explanations open: the oil, or the process. Close one of them by pouring a control oil in the same session: an oil you have already seen stay dry in this wax, at the same load, same addition temperature, same stir time, same jars and same cooling spot. If the control stays dry and the problem oil sweats beside it, the process is not the cause.

Record the addition temperature with a thermometer rather than by eye, stir for a fixed and timed period, and let all the candles cool in the same place away from fans and air-conditioning vents. If the control oil sweats too, stop thinking about the fragrance: fix the process first.

Read the example log row by row. Every rung is better than the one above it, which proves load matters for this oil. But even at 6%, which is the bottom of the stated range on many CSI fragrance pages, a film appears by day 14, while the control oil at 8% stays dry. The process is clean, and the oil cannot hold even a modest load in this wax. The observation days are a suggested starting schedule; your own may run longer.

Four stop rules

One rule is enough to start sourcing a replacement. Two or more and you should stop spending test candles on the current oil altogether.

Running the second-wax round properly

Choose a second wax you could realistically buy for production. For container candles in CSI's range that might be CSI 468 (coconut-soy, stated load up to 13% of wax weight) or Luxury Soy Wax Chunks (stated maximum 13% of wax weight). Convert before comparing: 13% of wax weight is 13 ÷ 113 = 11.50% of total, so a 7% or 8% total-basis test is comfortably inside both stated figures. Those figures are ceilings for typical oils, not promises for yours.

Pour three candles at 7% in the second wax and three in your current wax, plus a control in each. A 200 g candle at 7% is 186 g of wax and 14 g of oil, so six problem-oil candles use 84 g of oil. Observe on the same schedule. A one-kilogram pack of CSI 468 at ₹650.00 is enough wax for this round with some to spare.

What another round really costs you

The oil in each test round is the visible cost and it is small. In the example log, four rungs of three candles each use 3 × (20 + 16 + 14 + 12) = 186 g of oil. The bigger cost is time. A round that you observe to day 14 and then burn-test takes most of three weeks before you can act on it. Three rounds is two months in which the product is not on sale, the batch that failed is still sitting in quarantine, and the jars, wicks and wax for production are tied up waiting for an answer.

That is why the stop rules matter more than the ladder itself. A fourth round at 5% is tempting because the trend line is improving, but it pushes the oil below the range its page states, and the throw test usually closes that door anyway. Set the stop rules before the first round, write them on the test sheet, and hold yourself to them when the results arrive. Makers who decide the rules after seeing the results tend to keep finding reasons for one more round.

Choosing and testing the replacement

Pick candidates in the same scent family and test them beside the failing oil, not instead of it, so you have a direct comparison. Read each candidate's page for what is published: load range, flashpoint, vanillin where stated. For a vanilla, Vanilla Bean publishes 2.5% vanillin and a 203°C flashpoint; Dark Vanilla states vanillin is present without a percentage. Neither statement predicts stability in your wax. Only your test does.

A 100 g sample at 8% fills 100 ÷ 16 = 6 candles of 200 g, with 4 g left over. Two vanilla candidates at 100 g each cost ₹790.00 + ₹790.00 = ₹1,580.00 at list price. A 50 g pack at ₹410.00 fills three, which is enough for a first look but thin for a heat hold and burns. Run the candidates on the same ladder you used for the failing oil, starting at your target load, and include the control again.

When a candidate passes, it still needs wick testing at its load before you order production quantities. Then the old oil is retired from this wax, with a note in your formula records saying why, so nobody reintroduces it in six months.

Where this page fits

CSI already has guides that cover parts of this question. This page answers one sweating decision; these cover the neighbouring questions:

Three things to buy for this fix

Frequently asked questions