I’m Producing 5,000 Candles a Month — What Batch Records Should I Keep so I Can Reproduce Successful Production and Investigate Failures?

Why records change at 5,000 candles a month

At 5,000 candles of 200 g a month, you are handling 1,000 kg of scented wax: at 8% of total weight, 920 kg of wax and 80 kg of oil. In 25 kg batches that is 5,000 ÷ 125 = 40 batches a month, roughly two every working day. Nobody remembers the details of forty batches. The first time a complaint arrives about a candle made seven weeks ago, the only thing that can answer it is what was written down on that day.

Records at this scale have two jobs, and a good system serves both. The first is reproduction: when a batch performs beautifully, you want to know exactly what happened so you can do it again. The second is investigation: when a batch fails or a customer complains, you want to trace the candle back to its materials, its process and its people, and forward to every other customer who received the same batch.

Both jobs need the same thing: actual values, recorded at the time, linked by one code. A recipe card says what you intended. A batch record says what happened.

The six linked records

A small workshop does not need a quality system from a textbook. It needs six documents that point to each other.

The link that makes this work is the batch code printed on each candle's base label, and the receiving code written on each pack. CSI's product pages do not publish lot numbers, so the receiving code is yours to create: write it on the carton or bottle the day it arrives, and every batch sheet that uses it copies it across.

What goes on the batch record itself

The batch record carries most of the diagnostic value, so its fields deserve care. Group them in the order the work happens so the operator fills them in as they go rather than from memory at the end of the day.

Worked example: tracing one complaint in ten minutes

A customer writes that a Eucalyptus Spearmint candle bought last month barely scents the room. The label reads batch 26-08-14B. Here is what the linked records show.

The dispatch log shows batch 14B went to eleven customers. The batch record shows 23,010 g of CSI 464 from packs W-0812 and W-0813 and 2,004 g of oil from bottle F-0807, basis "8% of total", oil added at the wax maker's stated temperature, mixing as per the SOP. The pour log shows a 40-minute break at candle 96 with no reheat recorded, and candles 97–125 poured afterwards. The release record shows test candles taken only from positions 5 and 60.

The customer's candle carries number 112. The retained-sample register shows an end retain at candle 121 on shelf B3. Burned beside the beginning retain and a fresh reference, it is clearly weaker, with a rougher top. The cause is almost certainly the unrecorded cool pour after the break. The response writes itself: replace the customer's candle, check the other late-run candles from 14B still in stock, and add "reheat to validated window after any break" to the SOP.

Without the records, the same complaint would have led to a guess — perhaps a new oil order or a higher load. At 80 kg of oil a month, a guess like that is expensive: Eucalyptus Spearmint at ₹3,068.00 per kg means one extra percentage point across a month's 1,000 kg of scented wax adds roughly 10 kg of oil, about ₹30,680.00, to fix a problem that was never about the oil.

Using records to repeat a good batch

Investigation gets the attention, but reproduction is where records earn money every week. When a batch draws unusually good feedback, mark its record as a reference batch. Then compare it with the batches around it: what was different? Often it is something small — oil added a little cooler, pouring finished sooner after mixing, a particular wax pack, one operator's pour rhythm.

Once you have two or three reference batches, their actual figures become the target ranges on your SOP, replacing numbers that were originally guesses. That is how a workshop's process improves without anyone inventing a new recipe: the records quietly show which version of the same recipe works best.

Reference batches also make material changes safer. When a new wax pack or oil bottle arrives, the first batch that uses it can be compared directly with a reference batch record and its retained candles. If the numbers match and the retains agree, the new material has passed a real production test, not just a sniff at the bottle.

Keeping the system alive

Records only help if they are complete, and they stay complete only if they are easy. Use one template for every batch, keep blank sheets at the melter, and let the operator write readings as they are taken. A shared spreadsheet is fine; so is a bound book with numbered pages. What matters is that every batch has one record and every record is findable by batch code.

Review monthly. Look at fill weights, reject counts, temperature logs and complaints across the month's forty batches. Trends — a slow rise in rough tops, one operator's batches running light, one oil's batches drawing more complaints — show up in the records long before they become a failed batch. The review is also where you decide which successful batches become the reference you copy next.

Keep records at least as long as your retained samples, and longer for bestsellers. Filing invoices, receiving entries and the IFRA and MSDS documents CSI provides on WhatsApp request in the same place turns a pile of paper into a system you can actually search.

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