I Want to Reduce Candle Cost Without Changing the Customer Experience — Which Invisible Raw-Material Efficiencies Should I Investigate Before Changing the Visible Jar?

What stays fixed, and the worked candle

Before any lever, list the things your customer experiences and freeze them: the jar and lid they unbox, the label, the declared fill weight, the scent in the room (hot throw), the flame and melt pool, and the colour. Lane A cannot move any of these, because the materials are identical. Lane B can move all of them, which is why each lane-B change gets its own test and is kept only if the test shows no difference.

The worked candle is a coffee candle in the clear 200ml Salsa jar: a 160 g fill (a worked example; the page states no gram fill, and on the common 0.86–0.9 rule of thumb a 200 ml jar holds well under 200 g of wax), Roasted Coffee at 9% of total weight, Luxury Soy Wax CSI 464, a Cotton Braided wick, Brown Candle Dye Liquid at 0.3 ml, and a kraft Corrugated Box 11 x 11 x 8 cm. The worked reader makes 150 candles a month and buys small packs (also a worked example). Every unit price is CSI's; the quantities and pack choices are the worked example.

Wax, oil, dye and wick, the lines the customer never sees directly, are 70.4% of this candle. That is the territory of this page. The jar and box are the rest, and the jar is where a "visible" change would come later.

Lane A: same materials, better pack

Lane A is pure purchasing. The same jar, oil, wax and dye arrive in a different pack, so the candle is identical and there is nothing to test. The only questions are how much each pack size saves per candle, and how much cash it ties up at your volume.

The jar you already use, in a bigger pack

The clear 200ml Salsa is the only one of CSI's 200 ml jar listings that gets cheaper in a bigger pack: ₹40.12 a jar in 5s, ₹29.50 in 50s or 100s. At 150 candles a month a 50-pack lasts about 10 days, so the cash tied up is small. The one check: the page calls the jar "Upcycled", and an upcycled jar from a different batch can differ slightly in shape or glass. Before you buy 50, ask about batch consistency, and measure a sample of each new batch against the jars your wick was tested in. Why a few millimetres of diameter matter to the burn is the jar-dimension variation post.

Oil: most of the saving is already behind you at 100 g

Roasted Coffee falls from ₹6.21 a gram at 15 g to ₹3.90 at 100 g, and only to ₹3.74 at 1 kg. On a candle carrying 14.4 g of oil, that is ₹35.66 a candle between the 15 g and the 1 kg, but just ₹2.33 between the 100 g and the 1 kg; between the 500 g and the 1 kg it is ₹0.60. If you are still pouring from 15 g or 50 g bottles, this is your biggest lane-A lever by far. If you buy 100 g, it is a small one. CSI lists no fragrance pack above 1 kg; a 5 kg question is for WhatsApp and for the 5 kg fragrance pack post.

Wax and dye: real, but small

CSI 464 is ₹0.52 a gram in the 500 g bag and ₹0.51 from the 1 kg bag up; on 145.6 g that is ₹1.83 a candle, and nothing more at 5 kg or 10 kg. The brown dye drops from ₹15.73 a ml to ₹10.62: ₹1.53 a candle at 0.3 ml. Neither changes your life alone; together with the oil they are worth having because they cost nothing but a slightly bigger order.

Lane B: formula changes that must stay invisible

Lane B changes what is inside the candle. Each lever below is priced on the candle as it stands after lane A, so no saving is counted twice, and each counts only if a controlled test shows the customer cannot tell.

Load: the biggest lane-B lever, and the riskiest

Dropping from 9% to 8% removes 1.6 g of oil from a 160 g candle; the fill stays the same, so the same weight of wax replaces it. At the 1 kg oil price and the 1 kg wax price that is ₹5.17 a candle. Roasted Coffee's page states "candles (6-10%)", so 8% is inside the stated range, but a range is not a result: whether your candle still fills the room at 8% is only something your blind burn test can tell you. Which percentages to compare, and how, is the lower-load test post; the lowest load that still throws is the lowest effective load post. If 8% fails, the lever is worth nothing here and you stay at 9%.

Wick: you may be paying for a wick made for a wider candle

The worked reader uses Cotton Braided Wicks at ₹12.00. That page says they are "engineered for large-diameter candles exceeding 8 cm"; the Salsa page describes its jar as space for "votive or single-wick candles" and states no diameter. If your jar is narrower than 8 cm, a smaller wick may burn better as well as costing less: Eco C2 at ₹9.44 saves ₹2.56, C1 at ₹5.90 saves ₹6.10. Or it may not. Test the rungs side by side, at least three candles each, burned to the end, and change for burn first and money second. How many candles prove a wick is the three-candle wick approval post; the change-control maths of a cheaper wick is the ₹2 cheaper wick post.

Dye: over-dosing is common, under-dosing is visible

No CSI dye page states a dose, and many makers keep adding until the colour "looks right" and then never revisit it. Past a point, extra dye barely deepens the shade in an opaque soy wax; below it, the candle visibly pales. Pour colour chips at 0.3 ml and 0.2 ml per candle-equivalent, let them set and cure, and ask three people to pick the lighter one. If nobody can, 0.2 ml saves ₹1.06 a candle. If they can, the lever fails. Less dye is also a formula change for the wick in principle; the reverse case, more dye, is the darker-dye wick retest post.

Lane C: the yield, before the jar

One more invisible lever is not a material at all: candles you pour and cannot sell. At 5% rejects (worked example) the candle after both lanes costs ₹195.81 per saleable candle; at 2% it is ₹189.82, another ₹5.99, bigger than the wick and the dye together. Where those losses come from, and how to measure them, is the production-scale losses post; the full true-cost method is the true manufacturing cost post.

The flat-pack traps

Half of "buy bigger, pay less" is true at CSI and half is not. The ones that are not:

Roasted Coffee is a softer trap: the 1 kg is cheaper than the 500 g, but only by ₹0.60 a candle. Buy the size a month uses. If you want a lever on oil that is bigger than pack size, it is the load (lane B), and that needs a test, not a bigger bottle.

What it is worth, and what the tests cost

Annual value is saving per candle × candles a month × 12, at the worked reader's 150 a month. Lane A costs nothing but the bigger orders; lane B costs test candles at real CSI prices plus labour at ₹150.00 an hour (worked example; use your own).

The load ladder needs 115.2 g of oil and 1,324.8 g of wax for nine candles; buy the 1 kg of Roasted Coffee first (a lane-A move anyway) and 884.8 g is left for production. The wick bracket buys a C1 20-pack and a C2 10-pack (₹212.40), because the C1 10-pack is out of stock, and uses three of each. Every test candle is priced at full cost, and the payback is still short.

Sensitivity: at 500 candles a month lane A alone is worth ₹97,928.40 a year; at 50 a month, ₹9,792.84. At low volume, lane A is still worth doing because it costs nothing; lane B's tests take longer to pay back, so start with the load, where the money is.

What to buy for each lever

Five items behind the levers, chosen on page statements, price and stock on 4 October 2026. CSI publishes no throw or burn data, so nothing here is ranked on performance.

Skip for this decision: the 100-pack of boxes, the 1,000-pack of any tabbed wick, and the 10 kg wax bag bought for a discount: all flat per unit. And skip a different jar until both lanes are done; when you get there, the jar-change retest post lists what must be retested even with the same wax and oil.

Frequently asked questions