My Candle Passed Every Internal Test but Customers Are Reporting Tunnelling — How Do I Determine Whether the Formula Is Wrong or Customers Are Burning the Candle for Too Little Time?

Two explanations, one symptom

Soy container candles commonly keep a "memory" of their early burns: if the first few sessions end before the pool reaches the glass, the unmelted wax at the edge stays solid, the next pool forms inside the same ring, and each short session sinks the ring deeper. Many makers find the first burn matters most. That is the usage explanation, and it needs nothing wrong with the candle: a wick that reaches the edge of a 7 cm jar in about 2.8 hours (by the one-hour-per-2.5 cm rule of thumb) will still tunnel if it is only ever lit for one.

The formula explanation is that the wick gives too little heat for this jar, oil and wax in the conditions customers have, so even a patient first burn stops short. Your internal test may have passed because it ran in a warm workshop, with a freshly centred wick, or on a candle from a different pour. A test that passed is evidence about the candles and the room you tested; it is not proof about every candle in every home.

Both explanations produce the same photo. The way to separate them is not to argue with the customer or to re-wick on a hunch, but to reproduce their burn on your own candles and see which explanation the wax agrees with. That is the whole of this page: an intake that defines their burn, and a replica that performs it.

What each tunnelling report points to

Ranked in the order we would suspect them, most common first. The ranking is editorial judgement, not complaint statistics; CSI publishes none, and we don't know your customers. The last two columns tell you what would confirm or dismiss each cause.

Rows 1 and 5 end in a care card. Row 2 ends in a wick step. Row 3 ends in a centring check or a placement line. Row 4 is the two-room protocol's job, and row 6 is batch variation (3262). The replica burn below separates rows 1 and 2, which in our judgement are the two that matter most often and look most alike.

The customer-usage intake

Send the intake to every customer who reports tunnelling, and keep the answers in one sheet. Ask neutrally; "how long did you burn it?" gets a truer answer than "did you burn it long enough?". Record answers as given, including "don't know". We give no typical figures for customer burns because CSI has no such data, and yours will be more useful than anyone's average.

Three answers are enough to define the replica: the first-burn length (question 1), the usual session length (2) and the number of sessions (3). If reports disagree, replicate the shortest pattern that several customers describe, not the single most extreme one; that choice is judgement, and saying so in your notes keeps you honest later. Question 10 matters more than it looks: if a care card was in the box and customers still burned for an hour, a sticker alone won't change it, and the card's position or wording needs work.

The replica burn: customer-style against your protocol

Four candles from the batch the complaints came from, all at least as old as the customers' candles were when they noticed. If you kept no retained candles, pour four fresh from the same lots of wax, oil and wick; that is weaker evidence (a new pour is not the customer's batch) and your notes should say so.

Costs are worked examples and exclude labour, freight, packaging and GST. At 8% each candle is 12 g of oil and 138 g of wax, well inside the lower of Roasted Coconut's "up to 10%" and CSI 464's "Up to 10%". Pour at 80–85°C (the jar page's 85°C governs) and wait at least the oil page's 48–72 hour cure; many makers wait longer. If you retained candles, the cash cost of the whole test is nothing.

Verdicts from four candles

Two cautions. First, the formula verdict says "too little heat for real conditions", which is a wick question before anything else; more oil or a different wax changes several variables at once (the multi-problem hub sets the order). Second, a usage verdict is not permission to blame the customer. It means your candle needs a first burn it isn't getting, and your job is to make that instruction impossible to miss.

Writing the care card for a usage verdict

If the verdict is usage, the card has one job above all others: get the first burn long enough. Common practice for container candles is to let the pool reach the edge on the first burn, often 3–4 hours, and not to burn much beyond about 4 hours at a time. Put that first, in plain words, and put the card where it can't be missed: on the lid or the wick, not in the bottom of the box. Our suggested order (judgement): 1) "First burn: let the wax melt all the way to the edge, about 3–4 hours." 2) "Trim the wick before every burn." 3) "Keep away from fans, AC vents and open windows." 4) "Burn within sight; stop when about 1 cm of wax is left."

Then check it works. Hand two candles with the new card to people who haven't heard this conversation and ask afterwards how long they burned the first time. If they burned for an hour anyway, the card is in the wrong place or says too much. The recovery burn on R1 tells you whether a rescue line is honest; if R1 didn't recover, leave that line out rather than promise something the wax won't do.

Purchases that won't change a burn habit

If the replicas tunnel and your protocol candles don't, every raw-material purchase is a cost with no fix attached.

Products for the replica test

Five items in the order the verdicts need them: the card, the wick step, the jar and oil for fresh replicas, and a centring check. Chosen on page statements, price per unit and stock on 26 September 2026; CSI publishes no burn data, so nothing is ranked on performance. The oil table shows every size and 150 g candles at 8%.

Frequently asked questions