My Candle Formula Works at 100-unit Scale but Quality Becomes Inconsistent at 1,000 Units — Which Raw-Material and Process Controls Become Important as Production Scales?

What changes between 100 and 1,000 units

The worked example is a 150 g candle in the Clear Glass Jar with Golden Lid (150 gram), Luxury Soy Wax CSI 464 and Lavender Fragrance Oil at 8% of total weight: 12 g of oil and 138 g of wax per candle. At 100 units that is 1,200 g of oil and 13,800 g of wax; at 1,000 units, 12,000 g and 138,000 g. The recipe per candle is identical. What is not identical is how many times, from how many containers and by how many hands it is made.

Each row is a variable that was held constant at 100 units by the simple fact that everything happened once. At scale, holding it constant takes a deliberate control. Before building the plan, read your inconsistency against the pattern table below; it tells you which row is most likely to be yours. The ranking is our judgement, most common first, not measured frequency.

Rows 1 and 2 are process at scale; row 3 is material; rows 4 to 6 are organisation. If you already have many defects logged across lots, days and operators, the method for sorting them is the defect cross-tab; if your problem appeared the moment you went from a 1 kg pot to a 10 kg batch, the batch-size diagnosis is 3274. This page builds the controls that stop those patterns from appearing in the first place, and a single test to point you at the right row.

The control plan, tabled for 100 and 1,000 units

The table sets out each control at both scales. Frequencies and sample sizes are our judgement for a small Indian candle business, not an industry standard; the principle behind them is that every point where the recipe is made again gets its own record.

Three of these deserve a word. Temperature. The melter page states a display range of "60°C to 100°C" and "The ±10°C deviation". CSI 464's page gives "Pour Temperature 80°C – 90°C", and in the 150 g clear jar, whose page says "Safe for hot pour up to 85°C", the working window is 80–85°C, only five degrees wide. A display that may be ten degrees out cannot hold a five-degree window by itself; a thermometer you have checked against a second one can. A melter's temperature knob can make temperature feel handled; we would still read the wax. Retained samples. At 1,000 units they are the most reliable way to answer "was it always like this?" when a complaint arrives weeks later; one candle per melt, lidded and labelled with the melt number, is sixteen candles, and a sealed oil sample per lot is a few grams. Records. The batch sheet is the control that makes the others useful; the full template is the batch-sheet post, and the formulation side of scaling (heat time, mixing reach, pour duration) is the 1 kg to 25 kg checklist.

The two-scale test

The control plan prevents drift; this test finds the drift you already have. It separates the three big explanations (process at scale, material lot, and variation inside a melt) in one production day.

The ten test candles cost ₹2,365.13 of materials at ₹236.51 each (₹62.65 of Lavender at the 1 kg price, ₹70.02 of CSI 464 at the 10 kg price, ₹94.40 for the jar and ₹9.44 for a C2 wick, as the worked example's wick), excluding labour, freight, packaging and GST. Six of them come out of a production melt, so their real cost is the sale you give up. For the general version of this question in fragrance terms, why an oil that worked in a small test behaves differently in bulk, see the small-test-versus-bulk post.

Burn QC and release, by melt

At 100 units, a couple of burn-tested candles per pour day tell you about the run. At 1,000 units a failure is more often a failed melt than a failed product, so the sample has to reach every melt. The worked example's plan, all judgement: one retained candle and one QC burn candle per melt (sixteen of each) and three full burn-downs per run taken from the first, a middle and the last melt, stopping with about 1 cm of wax left, as is common practice. That is 35 candles, 3.5% of output, against 8 at 100 units (8.0%). The share falls with scale while the absolute number rises: ₹8,277.96 of materials at 1,000 units against ₹1,892.11 at 100. Budget for it; it is cheaper than a returned melt.

Release by melt. Each melt's candles sit on the rack, labelled with melt number and pour date, until that melt's QC candle has passed at the house cure age. Lavender's page states no cure time in the text we captured, so the house cure age comes from your own tests (many makers judge soy at day 7 or later); the method for standardising it across pour days is the cure-standard SOP. A melt whose thermometer reading fell outside the pages' window, like melts 7 and 13 in the chart, is held until its QC candle has been burned and passed; if it fails, the melt's candles are reworked or rejected, and the retained candle from the previous melt tells you whether the problem started there.

What not to buy to fix inconsistency

We sell the melter and we'd like you to buy it when you need it. If you are making 100 candles a month, you don't: a couple of pots and a thermometer are enough, and the ₹2,360.00 1-litre mini melter is a convenience, not a control. At 1,000 units the 10 kg melter earns its place, spread over one run at ₹21.24 a candle, as long as a checked thermometer and a batch sheet come with it.

Equipment and materials for the 1,000 run

Five items: the melter, the wax and the oil of the worked example, and the two tools for in-process checks. Chosen on page statements, price and stock on 26 September 2026; CSI publishes no batch-consistency or performance data, so nothing is ranked on performance. A second Pen Thermometer (₹354.00) to check the first is part of the temperature control; its page states no range or accuracy in the text we captured.

Frequently asked questions