How to Keep Fragrance Consistent Across Production Batches

What “consistent” means to a customer — and what they actually notice

A repeat customer does not compare your batches. They compare a memory of your candle with the candle in front of them, in a different room, on a different day, at a different cure age. That is an unfair test and you will never win it completely. What you can do is remove the variation that is genuinely yours, so that when someone says “this one is weaker”, you can answer from a record instead of a feeling.

Four things drift in a small Indian workshop, and they drift in a predictable order. The ratio drifts first, because a load is not a number you choose — it is the result of two separate weighings, and both of them move. The temperature drifts next, because a pot on a gas ring in April is not the same pot in a Mumbai monsoon kitchen in July. The cure age at sale drifts third, and it is the most invisible of the four: the batch you poured for a Diwali order and shipped on day 5 is chemically the same candle as the one you kept until day 18, and it does not smell the same in the jar. Only fourth does the oil itself come into it — a new pack, a different pack size, a bottle that has been open through two summers.

This page is a control system for the first three, plus the paperwork that lets you prove which one moved. It is deliberately not about mixing mechanics at scale, and it is not a cure guide; it is about holding one recipe still while you make it forty times.

Lock the ratio: your load is two weighings, not one number

Fragrance load in this series, and on CSI's product pages, is a percentage of wax weight: oil grams = wax grams × load. So 8% of 200 g of wax is 16.00 g of oil, and the poured candle weighs 216 g. If you use the other convention some makers prefer — percentage of the finished candle — the same candle reads as 7.41%, which is why two makers can describe the same recipe with two different numbers. Pick one convention, write it at the top of your batch sheet, and never mix the two.

Now look at what the scale does to it. A load is a ratio, so an error in either weighing moves it. Weigh 200 g of wax on a 5 g kitchen scale and you might really have 197.5 g or 202.5 g; weigh 16 g of oil on the same scale and you might really have 13.5 g or 18.5 g. Combine the worst cases and a recipe you believe is 8% is somewhere between 6.75% and 9.25% in the jar — a spread of 2.5 percentage points, which is larger than the difference between a starting load and a recommended one on the Luxury Soy Wax Chunks page (8% starting point, 10% recommended, 13% maximum).

The same arithmetic converts straight into rupees. At Lavender 1 kg pricing of ₹5.22/g, being 0.5 g heavy on every candle in a 25-candle batch spends an extra ₹65.26 of oil you did not intend to use — and being 0.5 g light gives 25 customers a candle you did not intend to sell. A whole percentage point of load on a 5 kg batch is 50 g of oil, ₹261.05. That is the real cost of a cheap scale: not the accident, but the accident repeated forty times a year in both directions.

Weigh for the batch, not for the candle

This is the single highest-return change in the whole page, and it costs nothing. If you weigh 16 g of oil into each of 25 jars, you make 25 weighings and inherit 25 errors, each of which is a large fraction of a small number. If you melt 5 kg of wax and weigh 400.0 g of oil into it once, the same ±0.5 g scale error is now ±0.13% of the dose instead of ±3.13%. The oil is ₹2,088.40 in one reading rather than ₹83.54 in twenty-five.

The batch method has a second benefit that matters more in a monsoon month than a dry one: every candle in the batch is drawn from the same homogeneous melt, so the variation within the batch collapses as well. A customer who buys two jars from the same order gets two jars that match. That is what makes gift sets and corporate orders defensible.

Lock the heat and the clock — by grade, not by habit

Add temperature is the variable most likely to move without anyone noticing, because nothing in the finished candle shouts about it. A pot left one minute longer on a hot April afternoon in Ahmedabad sits at a different temperature from the same pot on a wet July evening in Kochi, and “when it looks clear” is not a temperature. The fix is a thermometer and a written figure, and it takes about four seconds per batch.

The second half of the fix is grade discipline. CSI's soy and soy-blend grades publish quite different working figures, so a workshop that pours two grades needs two lines on the wall, not one. Where a page publishes nothing — Eco Soy CSI 400 does not publish a fragrance ceiling — write “not published” and ask on WhatsApp rather than borrowing a number from another grade.

Two practical notes for Indian workshops. First, the flash points printed on oil pages are transport, storage and handling properties — they are not the temperature at which fragrance “burns off”, and the 80–90°C add window sits well below most of them. But CSI's list does contain some low figures, and for those you should work at the cooler end of the window and not hold the melt hot: Neroli publishes 85°C, Palo Santo 93°C, Gardenia Tuberose 98°C. Second, a melter with a thermostat removes the single largest source of temperature drift in a home kitchen; the Mini Electric Wax Melter is ₹2,360.00 and the imported 10 kg unit is listed at ₹21,240.00. Neither is necessary to make consistent candles — a thermometer and attention will do it — but if your add temperature is the field that keeps changing on your sheet, that is where the money is worth spending.

Stirring is the cheapest lock of all. The Chunks page's instruction is slow stirring for two full minutes; slow matters because whipping air into soy changes the surface, and two minutes matters because dispersion is what binds oil into wax rather than leaving it to pool. Use Bamboo Stirring Sticks at ₹118.00 for five, or a CSI Pro Mixer at ₹349.00 on its lowest useful speed for larger melts, and time it with a phone rather than counting.

The batch sheet: eleven fields, one line per pour

A batch sheet is not paperwork for its own sake. It is the only tool that turns “this one smells weaker” into a question with an answer. Eleven fields fit on one line of a notebook or one row of a phone spreadsheet, and the whole entry takes under a minute if you write as you go rather than afterwards.

Two additions make the sheet worth keeping. Put the batch code on the candle — a base sticker or a line on the care label does it; Candle Care Stickers are ₹413.00 for 500, which is ₹0.83 per candle. And keep the oil pack label, or a photograph of it, stapled to the month's sheets. If a lot or batch number is printed on your pack, copy it into the oil field; if it is not, record the pack size and the date you opened it, and ask on WhatsApp when you need documentation tied to a specific consignment.

Resist the urge to design something elaborate. Brands that keep records for years keep short ones. Eleven fields, one line, one notebook that lives next to the melter, photographed to your phone once a week so that a spilt jug does not take your history with it.

Retained candles and the blind nose check

Keep two candles from every batch, labelled with the batch code, stored away from sunlight and heat. One is your reference for cold throw and appearance; the second is for burning when you need a hot-throw comparison. At the series unit that is ₹418.47 of materials from a ₹5,230.90 batch — about 8% of the pour — and it is the only way to answer a complaint with evidence. Over a year of monthly batches, twelve pairs is ₹5,021.66: less than one batch, spent on knowing whether you have a problem.

The check itself must be blind, because a nose that knows which jar is which will find what it expects. Line up the new batch and the retained candle from the previous batch with the labels covered by a third person. Sniff the older one first, then the newer, then the older again — the first impression of the first jar is always the strongest and it flatters whichever candle you smell first. Ask one question only: same or different. Not better or worse; not stronger. If two people out of three say “different”, you have something to investigate, and the batch sheets tell you where to look.

Timing matters as much as method. Do the cold check on the same day of cure every month — day 14 is a sensible anchor for soy container candles, and the Chunks page's own guidance of ideally 1–2 weeks sits inside it. Do the hot check on a cured candle, three hours of burn, in the same room with the fan off. And do all of it with a rested nose: three to five minutes outside the workshop first, no chai, no agarbatti burning next door.

When the oil is the variable — and how to tell honestly

Fragrance oils are manufactured products, and any perfumery house's output carries some variation between production runs — that is ordinary, and it is true of every supplier in the world, CSI included. What is not ordinary is deciding that the oil has changed while your own weighing, temperature and cure records are blank. So the honest order is: clean sheet first, then the oil.

Here is what CSI does and does not publish, plainly. Oil pages publish, where the manufacturer states them, a fragrance family or note description, a candle usage range, sometimes a flash point, sometimes a vanillin percentage, sometimes an IFRA certification claim and sometimes a discolouration note. They do not publish batch-to-batch variance figures, and neither this page nor any CSI page will give you one. What you can do is ask: WhatsApp +91 7397976926 for MSDS and IFRA documentation for a specific oil, and tell us which oil and pack you are asking about.

Two changes on your side often masquerade as an oil change. The first is a pack-size switch: moving from 100 gm bottles to a 1 kg pack changes nothing about the oil but changes everything about how you decant, how long the pack stays open and how much headspace sits above it. The second is vanillin-led colour drift, which is cosmetic and predictable. The Cinnamon Vanilla page states 2.5% vanillin and a moderate discolouration note; Vanilla Bean also states 2.5% and warns to test with lighter waxes; British Rose and Jasmine state 0% vanillin and no discolouration. A batch that has yellowed more than the last one is usually older, warmer or more brightly lit, not differently made.

A ten-minute consistency audit for one batch

The audit is arithmetic on your own sheet plus two sniffs. It catches the errors that are still cheap to catch: a mis-computed load before 25 candles cure at the wrong strength, a fill weight that drifted 15 g across the batch, a cure date that will have the order shipping on day 6. The materials at stake on a 50-candle festive run are ₹10,470.80; the audit costs ten minutes and no money.

One more honest line about cleaning, because it is where cross-contamination between scents hides. A jug that carried a gourmand oil and then a floral one will carry a little of the first into the second, and the effect is strongest with the heaviest bases. Wipe hot, then finish with Isopropyl Alcohol Spray at ₹383.00 for 150 ml or ₹520.00 for a litre, and if you run several scents a week, keep a dedicated jug for your vanilla-heavy oils. It is cheaper than discovering a hybrid scent in a corporate order.

Frequently asked questions