I Sell 500 Candles per Month Across Five SKUs — How Do I Forecast Raw-Material Consumption Separately for Each SKU?

Why 500 candles is five forecasts, not one

A business that sells 500 candles a month is tempted to forecast one number — 500 — and divide it across SKUs by habit. That works for wax only if every SKU is the same size, and it never works for oil, because each fragrance is its own stock line with its own pack sizes. In the example on this page, the three 200 g jars, the 85 g tin and the 50 g velvet jar use very different amounts of wax per candle, so the same 500 units can mean very different kilograms depending on the mix. And within the mix, one SKU is climbing while another is sliding — which a single total hides completely.

So the method is: forecast each SKU separately, convert each forecast through its own BOM, and only then add up the materials that are shared. This page does exactly that for five example SKUs — Vanilla Bean, British Rose and Woody Oud in 200 g White Matte jars, Jasmine in the 85 g Golden Aluminium Tin, and Blue Ocean in the 50 g Heart Cutout Velvet Jar. The two small containers are deliberately the two CSI containers with published fills, so their BOMs need no weighed estimate. The 200 g fill in the White Matte jar is the reader's own weighed fill; CSI does not publish one for that jar.

Step 1: a forecast per SKU, from your own history

Start with three months of units sold per SKU. The table below is the reader's example history; every month totals 500. Three simple forecasts are shown side by side — last month, the three-month average and last month plus half the three-month change (a gentle trend). None is "right"; the point is that they disagree per SKU while agreeing on the total.

Vanilla Bean has risen ten candles a month; Jasmine and Blue Ocean have slipped. The average forecast under-orders Vanilla Bean and over-orders Jasmine; the trend forecast leans further the other way. This page uses last month as the forecast — the reader's choice, and a reasonable one when the mix is drifting slowly. The difference it makes is concrete: on the trend forecast, Vanilla Bean would need 2,879 g of oil instead of 2,796 g, and Jasmine 636 g instead of 672 g. If your history has a festive spike, the growth post shows how to size an order for a rising line without stranding stock when it flattens.

Step 2: each SKU through its own BOM

Each SKU has its own bill of materials. At 8% of finished weight: a 200 g candle is 16 g oil and 184 g wax; the 85 g tin is 6.8 g oil and 78.2 g wax; the 50 g velvet jar is 4 g oil and 46 g wax. Wicks are the reader's burn-tested choices — Eco Thick C2 in the 200 g jars, Eco Thin C1 in the tin and the velvet jar. That is the reader's result, not a CSI sizing rule: CSI publishes no wick-size-per-diameter chart.

Then the allowances, once each and in order: 2 QC candles per SKU (whole candles), an example 3% reject allowance (divide by 0.97), 4% pour loss on wax and oil, and 2% breakage on the White Matte glass only. The tins are aluminium, so no breakage line; for the velvet jars use your own breakage figure if you have one — it is left at zero here.

Read the matrix by column. The oil column is five independent purchases. The wax column adds up to one: 82.83 kg. The wick column adds up to two, because two wicks are in use. The container column stays as three, because each container belongs to particular SKUs — though the three 200 g SKUs share one jar line. That pattern — demand per SKU, purchase per material — is what the 20-SKU wax post scales up and what the one-jar post and the three-wick post do for containers and wicks.

Step 3: round to packs and price the month

Here is a CSI price quirk worth knowing. Luxury Soy Wax Chunks are ₹299.00 for 500 g — ₹598.00 a kilo — and ₹2,995.00 for 5 kg, ₹599.00 a kilo. So the arithmetically cheapest way to cover 82,835 g is 166 × 500 g at ₹49,634.00. The practical cover is 16 × 5 kg + 3 × 1 kg at ₹49,717.00 — ₹83.00 more for the month, and a fraction of the bags to open. This page buys the practical cover and says so; the 10 kg bag (₹5,999.00, ₹599.90 a kilo) is dearer per kilo than two 5 kg bags and is not used.

Two oil lines show why per-SKU rounding matters. Vanilla Bean is the dearest oil in this range at ₹7.65/g at 1 kg, so it is ₹21,502.00 of the order on its own — 16.9% of the CSI subtotal for 32% of the units. British Rose is one of the oils that genuinely gets cheaper per gram as packs grow (₹3.42/g at 100 g, ₹3.07/g at 1 kg), so its line rounds to two 1 kg bottles. The small SKUs round to small packs: Blue Ocean's 224.6 g is covered by 4 × 50 g + 2 × 15 g, because its 50 g and 100 g are the same ₹4.94–₹4.95 a gram and a 500 g bottle would sit on the shelf for months.

At the order's own rates the material cost per candle is ₹313.32 for Vanilla Bean, ₹239.81 for British Rose, ₹263.24 for Woody Oud, ₹110.15 for the Jasmine tin and ₹98.88 for the Blue Ocean velvet jar — wax, oil, wick and container only; no stickers, labels, lids, labour, freight or GST.

Buy now, buy after testing

All five SKUs are established, so October's production materials can be bought now. The buy-after-testing line in a forecast-driven order is usually a mix change: a new container size, a new wick for an existing SKU, or a sixth fragrance. In this example the reader wants to try Woody Oud in the 85 g tin as well. That is a new SKU with a new BOM — tin, wick and oil load all need to be confirmed together — so it enters this month only as trial candles.

The production lines leave almost nothing spare — 1 tin, 0.6 g of Woody Oud, 1 Thin C1 wick — so the trial is bought as its own small lines: 2 × 15 g of Woody Oud for the 21.2 g three tins need, 1 × pack of 12 tins and a 20-pack of Thin C1 (the 10-pack was out of stock on the 4 October 2026 re-check). That is ₹664.44 against ₹1,26,923.74 of production materials. If the trial passes, November's forecast gets a sixth row and its own BOM; the spare tins and wicks from the trial packs go into it.

The CSI products in this five-SKU order

Where this fits in the procurement system

Per-SKU forecasting is step two of the pipeline on the cluster hub. If you would rather start from last month's actuals and a shelf count, units sold → monthly requirement is the simpler version. Once the forecast is in, the shared lines can be bought on a different rhythm from the oils — a shared wax and a slow oil deserve different safety stock — and the consolidated monthly order shows how to put all of it on one invoice.

Frequently asked questions