How Do I Document a Custom Candle Fragrance Formula so My Production Team Can Reproduce the Same Scent and Strength Every Time?

What a formula has to capture

Most makers' first formula is a line in a notebook: "cherry blossom 60, mahogany 25, cinnamon 15, 8%". It is enough for the person who wrote it, on the day they wrote it. It is not enough for anyone else. Which cherry blossom oil, from which supplier? Eight per cent of what: the wax, or the finished candle? Which wax, which wick, which jar, how long a cure before judging? Each of those changes what the customer smells. A usable formula answers all of them, in writing, so that a team member who has never met the scent can reproduce it.

The load convention deserves its own line because it is the most common source of silent drift. CSI's convention, used in every guide, is that load is a share of the finished candle weight: oil = W × L. A maker who adds 8% of the wax weight instead makes a true load of 7.41%, and a new team member following a card that just says "8%" can do either. 8% of the wax or 8% of the finished candle? explains the difference; the card below states the convention in words.

The formula card

Here is the card for the example formula, filled in. Blend grams are given per 1 kg of blend, so any batch can be scaled by multiplication. The candle section uses CSI's convention and shows the fixed-wax alternative, so nobody has to guess. Temperatures follow the wax maker: CSI's own guidance for CSI 464 is 80–90°C; for Luxury Soy Wax Chunks, follow the product page.

The third row is the one that catches people out. If the team always melts exactly 10 kg of wax, 8% of the finished candle needs 10,000 × 0.08 ÷ 0.92 = 869.57 g of blend, not 800 g. Put whichever row matches how your team actually works on the card, and state it in words. The card is also the place to record tolerances; as a working example, ±0.5% of each component's target in a premix and ±1% of the dose into wax. Weighing error and tolerances explains how to set them.

The batch record

Every wax batch gets its own record, filled in as the work happens. It refers to the card by name and version, captures every lot number, and records actual values next to targets. A good batch record takes a few minutes and saves hours when something goes wrong, because it narrows the question from "what could have happened?" to "what did happen?".

The batch number format links everything: formula code, version, date and sequence. A record like HB01-v1.2-260926-01 tells anyone which formula, which version and which day, and the same number goes on the retained candle, the QC sheet and, if you use them, the product labels. Standardising cure time across pour days and a repeatable mixing and curing protocol cover the process steps the record captures, and premixing a production concentrate has a draw log that attaches to it.

Version control and the change log

Formulas change: a component is replaced, a share is adjusted, the load is tuned, a new jar is introduced. Each of those changes the product. The rule is simple: any change to a component, share, load, wax, wick, jar or candle weight creates a new version, with a change-log entry saying what changed, why, what test supported it, and who approved it. The previous card is marked superseded and kept. Batch records always name the version they were made under.

Not every edit changes the scent; v1.2 above only clarified the card. It still gets a version, because the card the team uses changed. Minor clarifications and real formula changes can be distinguished in the "reason" column, but both are logged. When you replace a component, rerun the relevant tests before approval: which tests to repeat when switching suppliers and whether an almost identical oil needs retesting cover that.

Retained samples and QC release

Documents record what should have happened and what the team says happened. Retained samples let you check what actually happened, months later. Keep three kinds, each labelled with the formula, version and lot or batch number, stored closed, cool and away from light:

QC release is the last step of every batch record. At the same cure checkpoint used for approval, compare a candle from the batch with the approved reference, blind if possible: blotter match for the blend, colour match for the wax, and a burn score at least equal to the reference. Why cure age must match and making hot-throw testing objective cover the method. Only a released batch is sold. If a batch fails, the batch record and the samples are what let you find out why, and isolating a discolouring component and comparing a new supplier batch cover two common investigations.

Shopping list: the first documented production run

A first run under a new card should be big enough to prove the documents work and small enough to be affordable: here, 1 kg of blend, which at 8% scents 12.5 kg of candles, 62.5 candles of 200 g, with 11,500 g of wax. Each component is bought in the cheapest combination of listed in-stock packs that covers its grams; the small surplus becomes the retained component samples. Wicks and jars are for 62 full candles.

That is ₹301.26 per candle in materials across 62 candles, before labour, packaging, labels, freight and GST. The blend itself costs ₹81.70 of oil per candle at 100 g rates. Add a scale (₹354.00) and jug (₹531.00) if you do not have them. Labels, folders and printing for the documents are not CSI products; use whatever your team will actually keep up to date.

Frequently asked questions