How to Maintain Consistency Between Large Soy Wax Candle Batches

What actually drifts between one large batch and the next

A hundred-jar batch is not a fifty-jar batch done twice. It runs longer, so the wax has more time to cool between the first jar and the last; it uses more than one pack of wax and more than one bottle of oil, so it can straddle two lots of material; and it is poured by a person who is more tired at jar 90 than at jar 10. Those three things — time, lots and fatigue — account for most of what makers describe as “the wax has changed”.

The honest starting point is that some variation is not a fault. Soy is an agricultural product and two packs are not laboratory twins; ambient temperature in an Indian workshop swings 15°C between a February morning and a May afternoon; a jar that has been sitting in a hot delivery carton pours differently from one that has been on a shelf. A batch system does not eliminate variation. It records it, bounds it, and makes the day it steps outside the bound visible immediately instead of six weeks later in a customer message.

The batch record card: sixteen fields, one side of a page

A batch card is not paperwork for its own sake. It is the only thing that converts “the last one smelled stronger” into a comparison. Keep it to one side of a page so it actually gets filled in, and fill it in during the batch rather than from memory afterwards — a card written at the end of the day is a reconstruction, and reconstructions are confidently wrong. Paper on a clipboard beside the melter works better than a phone with wax on it; transcribe it into a spreadsheet the same evening if you want it searchable.

Seven numbers, and the band each one lives in

Every process control comes down to the same question: how far is this allowed to move before I stop and fix it? Answer it in advance, in writing, and the batch either passes or it does not. Answer it in the moment and you will always find a reason to carry on pouring. The bands below are CSI's working practice for soy container candles — they are a starting point tight enough to matter and loose enough to hit in a real workshop, not laboratory limits.

Two of those deserve a sentence more. The load convention is the most common invisible difference between two batches that look identical on paper: CSI product pages state the load as a percentage of wax weight, so 20 kg of wax at 8% takes 1,600 g of oil, and the finished 21.6 kg of candle is about 7.4% oil by total weight. Both numbers are defensible; only one of them is what your supplier's page means. Write “of wax weight” on the card and the ambiguity never costs you a batch. And cure is a tolerance, not a nicety — the same page that asks for 48 hours minimum says one to two weeks ideally, and 14 to 21 days is standard practice for soy container candles. Judging a batch early does not give you an early answer; it gives you a wrong one.

Lot tracking: knowing which pack and which bottle made which candle

Lot tracking sounds industrial and is mostly a discipline about opening things. The rule: finish one pack before you open the next, and write the pack's letter on the batch card. If a batch has to straddle two packs, split it into sub-lots at the pack boundary and record which jars came from which. Do the same with fragrance bottles — a 100-jar batch at 1,600 g of oil will cross a bottle boundary if you are buying in 500 g sizes, and that boundary is exactly where “the throw changed” lives.

Record whatever identifier is printed on the pack you receive. If there is none, create your own on arrival — a letter and the date, written on the bag in permanent marker before it goes on the shelf. It does not matter that the code is yours rather than the manufacturer's; it matters that two candles made from the same material carry the same reference. This is also why the receiving check and the batch card belong to the same system: the arrival code on the bag is what the batch card points at.

Colourants deserve the same treatment. One of our customers writes that the dyes “work with all kinds of waxes” and another that a little gives “light to dark depending upon the amount of dye you add” — which is exactly why the dose belongs on the card in drops or grams per kilogram rather than as “a squeeze”. The same applies to micas: record the shade and the batch you bought it in, because a colour difference between two production runs is the single most visible inconsistency a customer can spot on a shelf.

The quality set: three burned, one retained, per hundred

A quality system that never removes candles from the batch is not a quality system; it is an opinion. The working rule here is four candles per hundred: three pulled at random across the sub-lots and burned at day 14, and one labelled and shelved as a retained sample. At ₹185.51 of materials a candle — ₹119.80 of Luxury Soy Wax Chunks, ₹59.81 of Orchard Pear at 8% of wax weight, and a ₹5.90 Eco C1 wick — that is ₹742.03 against a batch carrying ₹18,550.80 of materials. Four per cent, to know rather than hope.

The retained sample is the part most workshops skip and the part that pays for itself first. Six months later, when a repeat customer says the scent is weaker than the candle they bought in September, you do not have to argue from memory. You take the retained jar from that batch, burn it beside a jar from the current one in the same room on the same evening, and you have an answer. Sometimes the answer is that the customer is right and something moved; more often it is that the customer's flat in October is a different room from their flat in March. Either way, you know.

Change control: one variable per batch, re-qualified before it ships

Every workshop changes things — a new grade, a bigger jar, a different wick, a supplier's new oil, a melter that holds temperature better. The discipline is not to stop changing; it is to change one thing at a time, on purpose, with the change written on the card before the batch starts. A batch with two changes in it teaches you nothing whichever way it turns out, and you will spend the next three batches trying to unpick which one did what.

The Indian year: the variables you record rather than control

Most process-control writing assumes a room with a thermostat. An Indian candle workshop is usually a room with a fan, and that changes the job: several of the seven variables above are not things you set, they are things you observe and allow for. Ambient temperature at pour is the obvious one. A batch poured at 26°C in January and a batch poured at 39°C in May will not set at the same rate, will not frost the same way and will not release from the jar wall identically — with the same wax, the same oil and the same pour temperature.

Monsoon adds humidity to the list, and the practical consequence is about materials rather than the pour: an open wax pack in a Mumbai or Kolkata July takes on surface moisture, which shows up as spitting when you melt and poorer adhesion at the glass. Since a hundred-jar batch opens 20 kg of wax, this is a batch-level problem, not a jar-level one. Seal packs after every use, keep them raised off the floor, and note the season on the card so that a July batch is compared with other July batches before it is compared with a December one.

Then there are the festival weeks themselves. Production pressure is a real variable: batches poured at 11 pm on a deadline drift more than batches poured on a Tuesday morning, because the pour temperature is watched less closely and the fill weights spread. If your card shows a scrap rate of 1% in September and 6% in the third week of October, that is not the wax. It is the calendar, and the fix is a bigger buffer in the plan rather than a different grade. With Diwali on 8 November 2026 and a 14 to 21 day cure, festive batches want to be poured by about mid-October — which means the last batch in which you can safely change anything is the one you are pouring now.

What the system costs, and what one uncontrolled batch costs

Process control feels expensive because its costs are visible and its benefits are invisible — you never see the batch you did not have to scrap. So put both sides on the same page. The recurring cost is the quality set: four candles per hundred, ₹742.03, or ₹7.42 a candle. The one-off cost is instruments: a Pen Thermometer at ₹354.00, a CSI Pro Mixer at ₹349.00 for a repeatable stir, a roll of Candle Care Stickers at ₹413.00 for 500 batch labels and a Silicon Measuring Beaker 250 ML at ₹600.00 — ₹1,716.00 together, which is ₹1.72 a candle across a thousand candles and nothing at all after that.

A second thermometer is worth the ₹354.00: one lives in the melt and one reads the room, and ₹708.00 for the pair removes the temptation to guess the ambient. If your batches are big enough that the melt sits waiting, a Mini Electric Wax Melter at ₹2,360.00 holds temperature far more steadily than a double boiler on a gas ring, and a melt that holds its temperature is the single easiest way to pull the first-to-last pour spread inside ±3°C. A Isopropyl Alcohol Spray at ₹383.00 and a Wick Centering Device at ₹118.00 remove two more sources of jar-to-jar variation for very little. A hot air gun is useful for reworking tops; stock on ours is currently down to a single unit, so check before you plan round it.

Against that: one uncontrolled 100-jar batch carries ₹18,550.80 of materials before jars, labels, packaging or your time, and 20 kg of wax alone is ₹11,998.00 at the 10 kg pack price with ₹7,476.00 of oil beside it. You do not need to scrap the whole batch for the arithmetic to work — ten jars is ₹1,855.08, which is more than twice what the quality set costs. And the loss that does not appear in any of these figures is the repeat customer who quietly does not come back because the second candle was not the first one.

One last honest note, against our own interest: none of this requires you to buy a different wax. If your batches are drifting, the first three things to fix are free — write the card, weigh the melt before calculating the oil, and log the pour temperature at the first and last jar of each sub-lot. Do those for three batches before you change a single material. Most consistency problems that arrive in our WhatsApp are solved by the card rather than by the catalogue, and we would rather tell you that than sell you a grade you did not need.

Frequently asked questions