My Candle Formula Works With Some Fragrances but Fails With Others — Why?

Five reasons one formula cannot fit every oil

Most makers build a formula on a first oil: a jar, a wax, a wick that gives a full pool, a load that smells good. Every new scent then goes into the same recipe, which is really a record of what the first oil needed. When the fifth oil needs something different, the formula seems to have "stopped working". It has not changed; the oil has. Five differences usually account for it. They are tendencies that vary by oil and wax, not rules, and CSI has no measured throw or colour data for any oil.

Why the load window matters most at the extremes

Of CSI's 102 fragrance oils, 48 pages state a candle load or range in their load wording, and two perfume-grade pages, Lavender Oud and Amber Oud Royal, list "Candles (Soy/Paraffin/Blends) 10%" in a usage table; the other 52 do not. Of those 48, three give a top figure below 10% (Coffee Vanilla Tobacco, Fresh Strawberries and one brand-inspired listing, each at 9%) and three allow more than 10% (Autumn to 11%, Cinnamon Vanilla and Japanese Cherry Blossom to 13%). At the bottom, 10 pages state a range starting at 8% and 3 at 7%. The consequence is simple arithmetic: a house load of 10% sits above three pages' stated maximum, a house load of 6% sits below the starting figure of 13 pages, and 8% sits inside every one of the 48 stated soy or candle windows (paraffin figures, often up to 6%, are separate). That does not make 8% the best load for every oil; it makes it the safest place for a house default, with 6% and 10% as per-oil exceptions tested within the ceiling. Whether every scent should carry the same load is argued in full in Should Every Fragrance in My Candle Collection Use the Same Fragrance Load?.

Remember the two-ceilings rule: the lower of the wax page's and the oil page's stated maximum governs. Luxury Soy Wax CSI 464 states up to 10%; Luxury Soy Wax Chunks and Premium Coconut Soy Wax CSI 468 state up to 13%. So Cinnamon Vanilla's "Up to 13% load for candles" is still a 10% oil in CSI 464, and Coffee Vanilla Tobacco's 9% governs in all three. These are formulation limits, not regulatory ones.

Is it the oil or the formula? Count the failures

Before you change anything, make the comparison fair. The failing oil and a passing oil must have been poured from the same wax, on the same day or within a day, with wicks from the same pack, into the same jar, at the same load in grams, and judged at the same cure age in the same closed room. If your "failure" is a candle poured last week against a favourite poured two months ago, you are comparing ages, not oils. Isolating a single weak candle is a separate job; here the question is narrower: across a range, where does the fault sit?

The third row is the one makers resist. If your formula suits a lavender and an oud but fails with a citrus, a vanilla and a floral, the likelier reading is not three "weak" oils but a wick that only just reaches the glass with the two easy ones. A formula that passes only with your most forgiving oils is a lucky one.

Is it actually the fragrance oil?

For the one oil that fails, three quick questions decide whether the oil is the cause. First, does its melt pool reach the glass by hour 3 to 4 in the same jar where a passing oil does? If not, it is a wick question for that oil. Second, is your load inside that oil's own page window as well as the wax's? If not, it is a load question. Third, with a full pool and a legal load, does a passing oil poured beside it fill the room while this one does not? Only then is the oil itself the likely cause, and only then is a replacement the right purchase. The evidence you need before declaring an oil weak is in How Do I Know If My Fragrance Oil Is Too Weak for Candle Making?.

Read the page before you pour: the signals that predict trouble

You cannot read hot throw off a page, but you can read what commonly makes an oil behave differently in a shared formula. The diagram applies six such signals to six CSI oils.

Stated load. Write the page's window on the SKU sheet next to your wax ceiling and use the lower top figure. Where the page states none, as on Mango Coconut (whose "Strong throw" is wording, not a load), work inside the wax's maximum, start at 8% and ask us on WhatsApp before going higher.

Notes structure. A pyramid with woods, musk, amber or vanilla at the base, like Fresh Strawberries ("vanilla sugar and creamy sandalwood base"), suggests an oil that keeps something in the pool as the burn goes on; a description ending on a citrus or fruit suggests one whose character may fade faster once lit (a tendency, not a rule).

Vanilla and vanillin. Sixteen CSI oil pages mention vanillin. Vanilla Bean and Cinnamon Vanilla state 2.5%; Gardenia states 0.2% and "minimal discoloration risk"; several, including Sandalwood, state 0%. Many more oils name vanilla in their notes without a figure. A 0% statement is the page's claim: vanillin is one common cause of yellowing, not the only one, so a white-wax range still needs a shelf check. Which oils commonly discolour is a question of its own in the discolouration cluster.

Cure and wick wording. At least 21 CSI oil pages state a soy cure, each of them 48–72 hours, and seven give a wick instruction, all of them some version of "test wick size to suit your container". Read the wick instruction for what it is: the supplier telling you the house wick may not suit this oil.

Some oils sit in two groups: Coffee Vanilla Tobacco is heavy, vanilla-bearing and window-limited at once, which is why it is the oil in our picks most likely to need a full set of exceptions. For a café or bakery range, where most oils are heavy and vanilla-bearing, the extra testing is covered in our guide to café and bakery candle collections.

House base plus per-SKU exceptions: how to run a multi-oil range

A range that works across many oils is not one formula. It is a fixed house base and a short exceptions sheet per scent. The base is what you never change for one oil, because changing it would upset every other oil in the range. The exceptions are what you are allowed to change per SKU, cheaply, without touching the rest.

Keep the exceptions sheet to one line per SKU: oil name and size bought, load in grams for your jar, wick, dye or jar colour, earliest judging day, and the date of the last test. The full SOP around that sheet, with forms and roles, is in How to Build a Fragrance Oil Testing SOP for a Candle Business; a programme for standardising 10 to 20 fragrances before bulk buying is in My Candle Business Needs Consistent Hot Throw Across 10–20 Fragrances — How Do I Standardise Fragrance Oil, Wax and Wick Testing Before Buying in Bulk?.

When to change the house base instead. If the audit shows most of the range needs the same exception, it is no longer an exception. When six of eight oils only reach the glass on a C2, make C2 the house wick and treat the two light oils as the exceptions. When half your range discolours, a tinted jar such as the Black Glass Jar with Black Lid (150 gram) may belong in the base (its listing title carries a "Deliveries Starting" note, so confirm dispatch on WhatsApp). Changing wax is the last resort for a running range; change the wick and the per-SKU load first.

The range audit: twelve candles, one weekend of pouring

The audit answers one question: which oils in my range need an exception, and which exception? It changes only one thing per pair, the wick, and holds everything else at the house base. Load is held at 8% because, among CSI pages whose load wording states one, 8% sits inside every soy or candle window, so no oil is tested above its page maximum. The worked example uses six oils you already own and two wicks each.

Figures exclude labour, freight, packaging and GST. A four-oil range is eight candles and the same two wick packs.

What to buy: a wick bracket and four oils that behave differently

The first card is the first fix to try when a single oil fails. The four oils are not a "best oils" list: they are chosen because their pages show four different behaviours a house formula has to accommodate (light top-led, heavy woody, vanilla-bearing, narrow load window), so a maker building or repairing a range can test the extremes before committing. Chosen on stated notes, stated load, stated vanillin, price per gram and stock at every size on 25 September 2026, not on throw, which CSI has not measured. Candle counts assume 150 g at 8%.

Getting a range consistent before Christmas, 25 December 2026

Christmas falls on 25 December 2026, 91 days after these prices were checked. That is enough time for one audit, one round of exceptions, a confirmation pour and a cured production run, if you start in the next fortnight. Dispatch times depend on your location; confirm them on WhatsApp before you promise a customer a date.

Buy production oil only for SKUs that passed. Oil per 8% candle at the 1 kg price: Citrus Lemon ₹33.98, Sandalwood ₹60.32, Vanilla Bean ₹91.75, Coffee Vanilla Tobacco ₹103.19. A kilogram makes 83 candles of 150 g at 8%. If you are unsure which oils deserve a kilogram at all, our guide to deciding which oils deserve 1 kg inventory covers that decision.

Where the rest of this cluster goes deeper

This page is the hub for CSI's commercial troubleshooting cluster. Each narrower question has its own page:

For how many florals, woods, fresh and gourmand oils a 10-scent range should test in the first place, see our 10-fragrance collection planning guide.

Frequently asked questions