At What Batch Size Should I Stop Assuming My Small-Batch Process Will Scale Linearly and Begin Validating Commercial Production Separately?

Why there is no universal kilogram figure

Makers often want a number: stop trusting simple multiplication above 5 kg, or 10 kg, or 20 kg. The honest answer is that the break depends on your equipment, your pace and your formula, not on a round figure. A workshop with a large single vessel may scale smoothly to a size at which another workshop, pouring from a small melter in several fills, has already changed its process twice.

What does transfer from one workshop to another is the set of triggers: the physical changes that mean your large batch is no longer a bigger copy of your small one. When any trigger is reached, stop assuming and start validating that batch size as its own process. The triggers below are suggested starting points to validate against your own records.

Six triggers that mean the process has changed

None of these triggers says the batch will fail. Each says the evidence from the small batch no longer covers it. That is the moment to treat the new size as a validation batch, sampled and held before release, rather than as routine production.

A worked progression: where one maker's triggers fired

Here is one maker growing an Eco Soy CSI 400 formula with Blue Ocean at 8% of total weight, poured into 200 g jars. Each row is the next batch size they planned. Their own pace, about one jar every 20 seconds including handling, is used for the pour window; substitute your own.

Weighing one real fill told this maker that the melter held roughly 8.5 kg of melted wax, consistent with the product page's 9.5 litres at about 0.9 g per ml. So 9 kg sat at the edge of one fill and 12 kg needed two. Notice that triggers fired well before the obvious one. The 6 kg batch already had a pour window twice as long as anything tested before. It happened to pass, but it passed as a validation batch, with retains, rather than being assumed.

At 12 kg, three triggers fired at once. That is the size at which the maker stopped scaling and started validating a two-melt process in its own right.

A lesson from a published instruction

Eco Soy CSI 400 comes with a clear instruction on its product page: add fragrance at 80–90°C and stir for 2 minutes. It is the only published mixing time in the range, and it is stated for that wax, not for any particular batch size.

At 1 kg, a two-minute stir reaches every part of a small vessel. At 12 kg across two melts, two minutes per melt may or may not be enough; the instruction cannot tell you, because it was never written to cover your vessel. Follow the addition temperature, since that is a property of the wax, and check it with a thermometer because the melter's display carries a stated ±10°C deviation. Treat the stir time at larger sizes as a starting point that your start, middle and end retains must confirm.

The same thinking applies to any instruction written for a single vessel. A fill line on a jug, a habit of stirring until the wax looks clear, or a memory of how long the last batch took are all small-batch rules. They are useful starting points, and none of them is evidence at a size you have not yet made.

What validating separately looks like

Validating a batch size separately does not mean a laboratory study. It means making the new size under a written plan, sampling it, and holding it until the evidence is in.

The cost of a validation batch is mostly time, because a passing batch is still sellable. The 12 kg batch in the example uses 11.04 kg of CSI 400 and 960 g of Blue Ocean. At ₹399.20 per kg from the 5 kg pack (₹1,996.00 ÷ 5), the wax is about ₹4,407.17, and the oil at ₹4,956.00 per kg is ₹4,757.76, about ₹9,164.93 in total. That is the amount a skipped validation puts on trust.

If you have been scaling on assumption until now

Many makers only ask this question after a batch has already drifted. If that is you, go back through your records and mark which of your past batches crossed a trigger. Any unsold stock from those batches deserves a quick check: pull a few candles from different positions, rank them against a fresh 1 kg control, and release only what matches. Anything that does not match should be held while you decide whether to rework, sell as seconds or write it off.

From here on, add the trigger checklist to your batch planning sheet. Before every batch that is larger than anything you have validated, run down the six triggers. If none fires, carry on. If one fires, that batch is a validation batch. It is a two-minute habit that replaces a guess with a rule.

Where this page fits

CSI already has guides that cover parts of this question. This page covers the failure investigation for the question above; these go deeper on the rest:

Three things to buy for this job

Frequently asked questions