The two prerequisites everyone skips

The failure mode, written out. A customer says a candle tunnelled. You decide the wick was probably it, switch from Thin (C1) to Thick (C2), and move on. Four weeks later there have been no further reports, so you conclude the wick was the cause. But you also changed fragrance lot in that window, the weather changed, and the complaining units were a small share of a batch that mostly sold without comment. One outcome, four candidate explanations, and a permanently changed bill of materials resting on the weakest of them. Almost every maker owns parts of the fix; almost none owns the two that turn a complaint into a question with an answer.

Neither is software and neither is expensive. The record is one sheet per batch, filled in during the pour — the discipline behind the complete raw material list, extended to lots rather than items. The samples are three finished candles in a labelled carton, costing 1% of a 300-candle batch. Start with the next pour rather than reconstructing history: a record written from memory is worse than none, because it looks authoritative and is not.

The seven stages, as procedures

Each stage states what is captured, who does it, what it looks like written down, and the decision that closes it. A stage that closes with no decision is a conversation, not a stage.

Counting complaints: rates, baselines and what moves them

Raw counts are almost useless, because they rise with everything — sales, batch size, channel, season. Count reports per hundred units shipped, four ways: per batch, per fragrance, per production week, overall. That gives you a baseline, which is what separates a signal from noise. Without one every complaint looks like a crisis, and a maker who treats it that way reformulates constantly and learns nothing.

Two of those rows use the same four reports and reach opposite conclusions, purely because the denominator differs. That is the argument for rates in one line. The fragrance cut finds something most often, because fragrance is usually the variable that moves between batches.

What the system costs, at CSI prices

The reference build behind every figure here: a 200 g candle at 16 g of fragrance oil and 184 g of wax, in a glass jar, with one wick and one wick sticker, priced at the largest in-stock pack.

Three retained samples cost 3 × ₹211.69 = ₹635.07, which out of a 300-candle batch is 1%. Re-pouring that batch costs 300 × ₹211.69 = ₹63,507.00 in materials alone, before labour, packaging, freight already paid on the units that shipped, and the selling time the re-pour eats. Quarantining it instead holds the same ₹63,507.00 still — nothing destroyed, but the working capital stops moving, which for most small makers is the more painful number.

A full quarter of retained samples, labelled, plus both instruments, is ₹8,358.72 — 13.2% of one scrapped batch. That is the entire commercial case. On a cheaper fragrance it falls further: swap Sandalwood for Citrus Lemon at ₹2,832.00/kg and the 16 g is ₹45.31, the candle ₹176.57, three retained samples ₹529.71. For where these costs sit inside a retail price, the 2026 pricing guide and the profitability breakdown use the same kind of build.

What cannot be looked up — and what this system will not do for you

Say this before you start, because it changes how you work. CSI publishes no wick sizing chart, no hot-throw rating, no heat-stability figure, no container-temperature figure and no burn data of any kind — nor a jar fill capacity for most jar families, and where a listing says "200 ml" that is a volume, not a 200 g fill, so pour one, weigh it, and write your own number down. The ceilings on three waxes are the only published formulation limits in the catalogue: Luxury Soy Chunks 13% of total wax weight, CSI 464 up to 10%, Coconut Soy CSI 468 up to 13%, the other thirteen publishing none, and all three set by the wax maker rather than by regulation.

So your own records are the only authority for almost every question this system asks. How your product behaves in the final third of its life, whether Thin (C1) or Thick (C2) belongs in your jar, how a scent reads in a particular room — none of it is looked up; it is established by burning your own units and writing down what happened. The corollary applies to every supplier, this one included: a performance claim without its test conditions is not information. Ask what was tested, in what vessel, at what load, for how long, and how it was judged.

And the honest limit. This system tells you what changed and what to do next. It does not make your product risk-free, and it does not turn a judgement into a calculation. It will not tell you whether a batch is fit to sell, because nothing on a supplier's page can: you are the person placing the product in a customer's home, and the decision and the liability both sit with you. Where a report involves damage, injury or anything with legal consequences, take qualified advice, keep every unit and record you have, and change nothing until you have it. What the seven stages give you is a case file worth having in that conversation. Buying-side reading: how to verify wax quality before you buy, premium versus cheap fragrance oils, and buying everything from one supplier, largely a traceability argument once you keep batch records.

Frequently asked questions