Should Every Commercial Candle Formula Have an Approved Raw-Material Specification Rather than Only a Recipe with Ingredient Names?

What a recipe of names actually allows

Most small candle businesses start with a recipe that fits on a sticky note. It names the materials and gives the weights, and while one person buys, pours and tests, that is enough, because the recipe lives in that person's head along with everything the note leaves out: which wax, which bag, which lavender, which wick, which jar. The trouble begins the day someone else buys the materials. A buyer holding a recipe of names is doing exactly what the recipe says when they pick the cheapest soy wax in stock. The recipe has no way of telling them that the candle was only ever tested with one of those waxes.

The table below takes the four lines of a typical recipe and lists what could legally fill each one from CSI's own catalogue alone — no other supplier involved. It is the simplest way to see the gap. Every reading is a real listing with a real price, and every one of them would satisfy the recipe as written.

The spread in the last row matters less than the spread in behaviour, which is invisible. The four waxes differ in stated fragrance ceiling (Chunks up to 13%, CSI 464 up to 10%, and no ceiling published in this data for Soy Pearl or CSI 468), and CSI publishes no melt point, hardness, frosting or burn data for any of them. That is not a gap in this page; it is the reason the page exists. When behaviour is not published, the only proof that a candle works is the test you ran — and that test is only valid for the exact materials it used. A recipe of names quietly lets the materials drift away from the test. The four sibling pages on single changes (wax, fragrance, wick and jar) show what each drift invalidates.

The specification line by line

An approved raw-material specification is not a laboratory document. For a small Indian candle business it is one page per SKU that answers five questions for every material: exactly what (listing title and handle, not a generic name), from whom (the approved source), in what form (pack sizes the buyer may order), how much (grams per candle, with the load convention written out) and how you check it on arrival (the incoming check against a retained sample). The last column is the one most makers skip, and it is the one that catches a problem before it reaches a production run.

Three lines in that template deserve a word. Approved pack sizes exist because the same listing bought in a different pack is still, most likely, the same material — but you are not told how lots are made or blended, so a new pack from a new order is treated as a new lot and gets the incoming check. Stated limit you rely on is where you write down what someone else has promised. For Luxury Soy Wax Chunks that is the wax maker's stated maximum of up to 13%; for CSI 464 it would be up to 10%; for every other wax in CSI's range no ceiling is published in this data, so the line would read “ask CSI on WhatsApp”. For wicks and jars the honest entry is “none published — your burn test”. Your measured values exists because CSI does not publish the White Matte jar's fill weight or internal diameter, so the only numbers that can go there are the ones you measure. The reproducibility record shows how to take and write those measurements so they survive staff changes.

What a silent substitution does to your test file

The retest matrix below uses the vocabulary that runs through this whole series: KEEP (the earlier result still stands), CONFIRM (a short side-by-side check against the control), REPEAT (a full retest) and NEW (a test you never needed before). Each row is a substitution that a recipe of names would allow without anyone noticing. Read across to see what it does to the evidence behind the approved candle.

Two patterns stand out. First, the same-listing rows are never KEEP across the board. CSI does not publish lot numbers or lot-to-lot consistency, so even an identical reorder earns a short CONFIRM on the attributes most likely to drift — burn and throw — which is a smell check, a look at a melted sample and, for a bestseller, one side-by-side candle. Second, a wax substitution empties the file. Every column becomes REPEAT, and one becomes NEW when the replacement has no published ceiling. That is the whole case for a specification in one row: a word like “soy” lets a buyer make the one change that costs you the most retesting. The wax-supplier hub and the fragrance-replacement hub walk through those full retests.

Qualifying the spec: prove it reproduces the approved candle

Writing a spec for a candle you already sell is not paperwork alone. It makes a claim — “these exact materials, in these amounts, make the candle our customers know” — and a claim like that should be checked once. The check is small: make candles strictly from the spec, burn them beside a control, and see whether anything differs. If you have retained candles from an earlier approved batch, they are the control. If you do not, the six-candle kit below lets you make both sides fresh, which still proves the spec is complete: any instruction you forgot to write down will show up as a difference between the person who wrote the spec pouring and a second person pouring from the page alone.

Safety lines have no numeric targets here because CSI publishes no flame-height or jar-temperature limit, and this page will not invent one. The acceptance rule is comparative: no worse than the approved control on every safety line, and within your jar supplier's guidance and the recognised candle fire-safety standard you work to (ASTM F2417 is a commonly cited one). Consumption in grams per hour is weight lost divided by burn hours — keep it, because it becomes the fingerprint of the candle in every later comparison.

The priced qualification kit

Six candles — three to the spec, three control — at 200 g and 8% need 96 g of oil and 1,104 g of wax. The kit uses only listed, in-stock CSI sizes. If you have retained control candles, halve the wax and oil and buy one pack of jars.

That is ₹367.50 per test candle, higher than the ₹270.45 production cost because test packs are small and the jar pack rounds up. Set it against what the spec protects. A 500-candle run of this SKU carries ₹1,35,226.00 of materials alone at listed prices, before labour, packaging, freight and GST. One uncontrolled wax swap that fails in customers' homes puts that at risk. Writing the spec takes time too — as an example, 3 hours at ₹300.00 an hour is ₹900.00; replace both with your own figures.

The decision rule: locked, alternate, or ask first

Give every material line one of three statuses. Locked means one exact listing and the approved pack sizes; the buyer orders it and runs the incoming check. Approved alternate means a second listing that has already passed a full qualification against the control — the buyer may switch to it when the primary is out of stock, and records the switch on the batch record. Ask first covers everything else: any material not on the spec, any new supplier, any pack size not listed. Nothing moves from ask-first to alternate without evidence. the approved-supplier list is where the alternates live, and pre-qualifying backups explains when it pays to build them before a stock-out forces you.

Releasing and maintaining the spec

Once the qualification burns pass, the spec becomes a controlled document. Give it a version number and an approval date, keep one master copy, and write the version on every batch record that uses it. The next time anything on it changes — a new pack size, a new alternate, a change in grams per candle — the change goes through a change request, the retest matrix decides what must be checked, and a new version is issued only when the evidence is filed. The simple change-control form is designed to sit next to the spec, and the full change-control system shows how the pieces connect at larger scale.

Where a business has a procurement person who is rewarded for saving money, the spec is also a protection for them. It turns an unclear instruction into a clear one, and it moves the decision about equivalence to the person who owns the test evidence. The procurement-freedom post looks at why “buy the equivalent” is the phrase that causes most uncontrolled changes, and the 20-SKU triage post shows how to decide which SKUs need a spec first.

Frequently asked questions