I Need to Change Wax and Fragrance Simultaneously Because Both Suppliers Are Out of Stock — How Should I Design Tests so I Can Still Understand the New System?

The problem with changing two things because you must

The textbook advice — change one variable at a time — assumes you control the timing. A double stock-out takes that away: the wax and the fragrance you use are both unavailable, orders are waiting, and there is no time for two sequential test rounds. The tempting shortcut is to pour the new wax with the new oil, burn a few, and ship if they look fine. That approves one new candle and teaches you nothing about either change. When one old supplier comes back next month, you will not know whether you can return to it without testing again, and if the new candle develops a problem in storage, you will not know which half caused it.

There is a better design that costs one round of testing, not two. Pour all four combinations of old and new at the same time — a 2 × 2 factorial with the old candle as its control — and you learn the effect of each change on its own, the effect of the combination, and you end up with up to three approved recipes instead of one. The catch is that it needs a small amount of the old wax and the old oil, and that stock has to be reserved before production uses it up. The one-variable-at-a-time post explains why attribution matters; this page is how to keep it when the calendar will not cooperate.

Step 1: reserve the old stock

The design has four cells, and two of them use each old material. Old wax goes into cell A (the control) and cell C (new oil in old wax). Old oil goes into cell A and cell B (new wax with old oil). With four candles per cell — three to burn and one to keep unlit for storage checks, a working practice — and a 200 g candle at 8%, the reserve is 8 × 184 g = 1,472 g of old wax and 8 × 16 g = 128 g of old oil, plus a 5% pour-loss allowance (a labelled working assumption): 1,545.6 g of wax and 134.4 g of oil.

If your shelf holds less than the top row needs, step down. Two burn candles per cell instead of three still shows a clear pattern, though one odd candle then carries more weight. Moving the whole design to CSI's Golden Aluminium Tin, which has a published 85 g fill, cuts the old-stock need by more than half — but the tin is not your SKU, so a tin result tells you about the wax and oil, not about your jar's burn. If you go that route, confirm the winning recipe in your real jar before release. If there is no old stock at all, cells A, B and C are impossible: the best reference is a retained finished candle from the last good batch, and the new system is treated as NEW in every line.

Step 2: the four cells

Everything else is frozen: the same jar from one carton, the same wick, the same load, the same fill weight, poured on the same day, cured on the same shelf with positions mixed, burned in the same room. Each wax is melted and poured according to its maker's instructions. The load stays at your approved 8% in every cell even if you suspect the new oil would do better at another level — load optimisation is a third change, and it comes after this test on the winning cell.

Why not just test D against a retained old candle? Because that is the all-at-once design in disguise: it approves D and nothing else. The factorial costs 16 candles instead of about 8, and in return it gives you the wax effect, the oil effect, whether the two interact, and cells B and C as ready-made fallbacks. When a stock-out is caused by two suppliers at once, the chance that one of them comes back first is high — and the factorial means you can switch back to it without another test round.

Step 3: scoring and the effect arithmetic

The three effect columns turn sixteen candles into three plain statements. The wax effect compares the two new-wax cells with the two old-wax cells, so it averages over both oils. The oil effect does the same for the oil. The interaction asks whether D is different from what B and C together would predict; a large interaction means the new wax and new oil behave differently together than apart, which is exactly the situation the two-changes-failing post is written for. For safety lines, do not rely on averages alone: look at every individual candle.

CSI publishes no numeric flame-height or jar-temperature limit. "No worse than A" is judged within your jar supplier's guidance and the recognised candle fire-safety standard you work to — ASTM F2417 is commonly cited in the US. CSI publishes no burn-test, throw or compatibility data for Natural Soy Pearl Wax or British Rose, and nothing on this page predicts how either will score; the cells decide.

The priced kit

The worked example is a 200 g rose candle in a White Matte Glass Jar on an Eco Thin C1 wick, where both the maker's old wax and old rose oil are unavailable. The candidates are Natural Soy Pearl Wax and British Rose Fragrance Oil. The kit buys the new materials for cells B, C and D, plus jars and wicks for all 16 candles; the old wax and oil come from the reserve. No fragrance ceiling is published in this data for Natural Soy Pearl Wax — check the product page or ask CSI on WhatsApp before assuming your 8% is inside it.

₹2,988.60. The oil is bought as 1 × 100 g + 1 × 50 g because the test needs 134.4 g and British Rose's per-gram price only falls meaningfully above 100 g (₹3.42/g at 100 g, ₹3.19/g at 500 g, ₹3.07/g at 1 kg). If D is approved, production economics look different: at the 5 kg Soy Pearl rate (₹424.80 a kilo — the one wax price break in this set) and the 1 kg British Rose rate, the new 200 g candle is ₹203.95 in materials (wax, oil, ₹5.90 wick, ₹70.80 jar), before labour, packaging, freight and GST.

Release, and what to record

Each approved cell becomes its own recipe on the SKU's specification, with the date and the test reference: "B — Soy Pearl with old rose oil, approved", "C — old wax with British Rose, approved", "D — Soy Pearl with British Rose, approved". The approved-supplier list gets the new materials, and the old materials stay on it as approved for the cells that used them. That is the real prize of the factorial: when one old supplier comes back, purchasing can buy from it immediately and production can use the matching recipe without a new test.

Two final points. First, the factorial is only as good as its control, so the day a stock-out is announced is the day to reserve old stock — before anyone decides anything else. Second, the best way to never need this page is to hold pre-qualified backups: a second wax and a second oil for each bestseller, already tested against your current candle. The backup-fragrance post, the pre-qualified-supplier post and the change-control hub show how to build that in calm weather.

Frequently asked questions