I'm Making Different Candle Fragrances in the Same Soy Wax — Should Every Scent Automatically Use the Same 8% Fragrance Load?

Why one load does not fit every scent

A fixed percentage fixes the weight of oil in the candle, and nothing else. At 8%, a 200 g candle holds 16 g of oil and 184 g of wax whether the oil is a sharp lemon or a dense sandalwood. That is the convention CSI uses in every guide: load is a share of the finished candle weight, so oil = total weight × load. If you add oil as a percentage of the wax instead, the true load comes out lower; the 1 kg batch question works through both conventions. Whichever you use, the weight is identical across scents. What differs is what that weight does once it is in the wax.

Candle makers and perfumers commonly describe four mechanisms that make two oils at the same load perform differently. None of them is a CSI measurement. We do not publish throw, dosage or composition data for any oil, and nobody can tell you an oil's best load from its name. These mechanisms explain why you need to test, not what the test will show.

Put those together and the picture is clear. A blanket load treats every oil as if it were the same mixture. For some it will be too low, and the candle will be weaker than it could be. For others it will be more than the candle can use, which costs money and can make the burn worse. The 8% candle that beat the 10% and the candle that burned better at 8% show how a higher load can reduce performance rather than add to it.

What a blanket 8% really costs

Here are the six oils used in this page's grid, priced at their 100 g rate so you can compare them on the size most small producers buy. Each row shows the oil cost inside one 200 g candle at 6%, 8% and 10%, and the last column shows how much the load decision is worth on a single candle.

The spread inside each row is the money a per-scent load can save or spend. For Lavender it is ₹43.20 a candle between 6% and 10%. For Citrus Lemon it is ₹24.00. Those look small until you multiply by a production run. The complete material cost of a Lavender candle at 8% — ₹86.40 of oil, ₹110.22 of Chunks, a ₹5.90 Eco Thin C1 wick and a ₹70.80 White Matte Glass Jar — is ₹273.32, excluding labour, packaging, freight and GST. Oil is roughly a third of it.

Now imagine your grid came back with different answers for different scents. The table below is a hypothetical outcome, used only to show the arithmetic: we have not tested these oils and are not claiming these loads. It compares running four scents at a blanket 8% with running each at the load its (imagined) grid picked, for 1,000 × 200 g candles of each, oil priced at the 1 kg rate.

In this example the per-scent plan costs ₹12,254.00 less on oil across 4,000 candles — and that is not the main benefit. The bigger gain is that Citrus Lemon, the scent that needed more, now gets it, and the two that passed at 7% stop carrying oil they did not use. The per-scent route is about putting oil where it earns its place. For the full money-saving version of this idea, see finding the lowest load that still throws.

The 18-candle test grid

The grid below tests six scents at three loads in one pouring session. It is deliberately small. Every candle uses the same wax, jar, wick, fill weight, pouring temperature and cure, so the only difference along a row is the fragrance percentage and the only difference down a column is the scent. If you change anything else between candles, you will not know what caused the result.

Run the grid and you will usually see three kinds of row. Some scents pass at every load, which means the lowest load is the right one. Some pass only at the top, which tells you the scent needs more oil or a better-matched wick. Some fail everywhere, and for those extra oil is not the answer. Weak at 6% — raise it or fix the wick? and strong cold throw, weak hot throw cover those failures.

Six oils for the grid, priced

The six oils below were chosen to spread the grid across families and price tiers: a fresh citrus, a herbal, a rose, a white floral, a gourmand and a wood. All six are in stock at 50 g and sit high in the collection's best-selling sort on the pull date. That is the only reason they are here. We have not published throw measurements for any oil, and nothing below predicts which load each will need.

That is ₹297.43 per test candle, all materials included. Each test candle costs more than a production candle because you are buying oil at 50 g and jars in fives. The point is not to make sellable stock. It is to avoid pouring 500 candles at a load nobody tested. If you already own jars from earlier batches, reuse them for tests and the cost drops by ₹1,416.00.

When one house load is fine

None of this means a single load is wrong. Plenty of small producers run one load across most of their range, and there are good reasons to. One number is easier to train a helper on, easier to check, and harder to get wrong on a busy pouring day. Every extra load in your production sheet is another chance to weigh 20 g of oil into a candle meant for 16 g.

The two-load compromise is often the best production answer. Group your scents into a "standard" and a "lifted" load after the grid, write both on the batch sheet, and colour-code the jugs or labels. You get most of the benefit of per-scent loads without a production sheet full of different numbers.

Record everything. For each scent, write the wax, wick, jar, fill weight, load, addition temperature, stirring time, cure age at test, and the result. When you later reorder the oil — whether 100 g or a bulk order sized from your tested load — that record is what lets you reproduce the candle.

If you are launching many scents at once, eighteen candles for six scents scales badly: twenty scents would need sixty candles. The faster method for twenty scents screens everything at one load first and ladders only the finalists. If one scent is clearly underperforming next to the others, the lavender-versus-vanilla guide walks through the diagnosis before you touch the load.

Frequently asked questions