Custom Handbag Color Matching: Pantone to Production Reality

How close will production get to the Pantone reference on your screen? Honest answer for leather and coated material: Delta E 2.0 or better against an approved physical standard, never against a picture. Colour is approved against a swatch held under a defined light source, not against a hex value, because a hex value has no thickness, no grain and no dye uptake — and yet nine colour disputes in ten start with somebody approving a screenshot. The sequence that works is a Pantone TCX reference, a physical lab dip at least 10 x 10 cm, approval under D65 with a TL84 secondary check for metamerism, and a tolerance stated as a number that both sides can measure later. Ask for those four and the difference between sample and bulk becomes a specification rather than an argument.

The four things a colour approval needs

What Happened: An Approved Screen Colour and 500 Bags

A Melbourne accessories label came to us with a finished tote in three colours and a problem. One colourway had been approved from a photograph sent on a Thursday, matched to a Pantone TCX reference the designer had picked off the chip rack, and signed off by email that evening without anybody holding a physical swatch. Sampling proceeded at the standard 6-10 working days and produced exactly what had been asked for: a correct match to the reference, viewed under daylight bulbs in Fujian. Then it traveled. By the time the bag sat in a shop window in Melbourne under warm retail lighting it read noticeably greener than the picture that had launched it, and 500 units of it read the same way.

Nothing had gone wrong in the conventional sense. The dye recipe was right, the operator was competent, and every panel was within tolerance of the standard that had been sent. What failed was the step before any of that: the standard itself was a hex value and a memory. The re-run cost more than twice what a physical approval would have cost, in dye-lot minimums rather than in labour, because each colourway is its own dye lot and there is no economical way to make a fifth-and-a-half of one. The rest of this article is the sequence that should have been used, and it takes about four days longer than the one that failed.

Why the Screen Was Always Going to Disagree

Three things separate a screen from a dyed hide, and none of them are fixable with a better monitor. First, a display emits light in three narrow bands while a dyed surface reflects light across the whole spectrum, so two surfaces can be indistinguishable in one and clearly different in the other — this is metamerism, and it is quantifiable as an index rather than as an opinion. Second, leather accepts dye unevenly by fibre density, so a shoulder panel and a belly panel take up the same recipe differently. Third, finish matters: a matte topcoat absorbs, a gloss topcoat reflects, and the same dye under those two finishes reads as two colours.

• Metamerism. Measured against two light sources — D65 for daylight, TL84 or CWF for retail fluorescent — and expressed as an index the customer accepts or rejects. Below 1.0 is a practical pass for handbags.
• Sheen and grain. A 12 percent gloss difference reads as a colour difference even on a spectrophotometer that reports the two as identical.
• Camera white balance. Photographs of samples are nearly always taken under mixed light with automatic balance, which reliably moves warm tones by several Delta E units before anyone sees them.
• Substrate colour. A dyed hide under a pale correction coat versus the same hide under a pigmented finish will not meet even though both were mixed from one recipe.

The practical consequence is simple and worth stating plainly: colour disputes are almost never disputes about dye, they are disputes about which object was the standard. Once both sides agree the standard is a physical swatch held under D65 with a TL84 cross-check, the argument ends before it starts, because there is now one object both parties can put under a lens.

The Three Documents That Replace a Screenshot

A screenshot is replaced by three things, and each is cheap enough that there is no reason to approve colour without them. Ask for them in the first email and they arrive with the quotation rather than being extracted after a disagreement. Every one of them is a physical object or a number, which is the whole point.

1. A named Pantone TCX reference, written into the tech pack as a number rather than described as 'caramel' or 'deep olive'. Descriptive colour names are the single largest source of avoidable revision.
2. A physical lab dip of at least 10 x 10 cm, dated, signed by both parties, with one half retained on file and the other kept by the brand. Small swatches under 5 cm cannot be measured reliably with most apertures.
3. A tolerance statement in Delta E, together with the light source used and the aperture size, so the value can be reproduced later rather than being a number recorded once and argued over twice.

Those three documents are also what make a repeat order work. The second run does not restart from a search; it restarts from a retained half-swatch and a recorded tolerance, which is why colour is usually the first thing to get faster on a reorder. There is more on how that compounds across a season in the real timeline from brief to delivery, where the colour step sits between day 4 and day 9 of a typical calendar.

How Dye Lots Work: Where Colour Variation Comes From

Every colourway is a dye lot or a coating run, and that has two consequences that shape pricing and acceptance. A lot has a minimum quantity regardless of how many units are cut from it, which is the mechanical reason MOQ is 500 pcs per style / colourway rather than 500 across a mixed order. And variation exists within a lot as well as between lots: the beginning, middle and end of a 300-metre coating run drift, and hides from one delivery are not identical to the next even at the same grade.

• Within a lot. Practically held to plus or minus 0.5 Delta E across the length of the run, managed by cutting components from sequenced lengths in the order they come off the line.
• Between lots. Expect 0.8-1.5 Delta E even with a retained recipe, which is precisely why the retained half-swatch matters more than the formula.
• Natural hides. Grain density varies across a single hide, so panel-to-panel tolerance on full grain should be stated as Delta E 2.0 rather than the tighter figure used for coated material.
• Batch tracking. Every carton of finished goods should carry the lot number so a later claim about one colour can be matched to the material used rather than guessed at.

This is also why matching hardware colour to body colour is a separate and harder problem. A zinc alloy part finished in light gold is a plating bath, not a dye, and the two processes do not share a tolerance. Where a logo bar and a body panel must read as one colour, specify them as a set against a single standard, expect three attempts to get them together, and accept a slightly looser tolerance — usually Delta E 2.5 on the metal — rather than forcing the dye.

Reading Delta E Without Being Misled by It

Delta E gets quoted as though it were a physical quantity with one meaning. It is a difference between two measurements, and the formula used decides what the number means — which is why two people can measure the same pair of swatches and disagree by half a unit. Specify the tolerance, the formula and the light source together, or specify none of them and accept that the number cannot be enforced later.

One more trap: a tolerance stated for the sample but not the bulk is a tolerance that will be applied to neither. Put it on the purchase order with the light source next to it, and require that the pre-shipment inspection — AQL 2.5 under ISO 2859-1 — includes a colour check against the retained swatch as a named defect class. Colour that is not a named defect class is uncounted: unless it carries an accept and a reject number, it is invisible to the inspector.

Material Choices Change Colour Uptake More Than the Recipe

The most useful thing a designer can learn about colour is that the material decides how a recipe looks more than the recipe does. A colour developed on full-grain calf will not read the same on corrected grain, and neither will read the same on 600D polyester with a 0.35 mm PU coating. Two consequences follow: develop colour on the material you will actually ship, and never ask an existing colour reference to be matched onto a new substrate without a fresh lab dip.

• Full grain aniline. Richest depth, most panel-to-panel variation, and the most sensitive to sunlight over the life of the bag.
• Corrected grain with a pigmented finish. Most consistent colour across a hide, slightly flatter appearance, best choice where exact matching is required.
• Coated synthetic. Most predictable colour of the three, excellent rub fastness, weakest depth perception under directional light.
• Textile with print or transfer. Three-dimensional artwork achievable, but colour matching is to a printed standard rather than to a dyed one, so a different tolerance applies.

The custom versus private label decision touches the same point from another direction: choosing a custom dye development buys control and costs a lot minimum, whereas working within an existing stock colour range skips the whole exercise and delivers consistency from day one. Neither is wrong; they solve different problems.

Colour Fastness: The Standard Nobody Quotes Until Something Fades

Matching is about day one. Fastness is about month nine, and it is the part of the specification most often left out because it does not appear in product photography. Three test families cover almost everything relevant to a handbag, and quoting them by standard number costs nothing at brief stage while protecting the whole first order.

Light fastness

ISO 105-B02, grade 4 minimum for any colour carried in a window or near glass

Rubbing, dry and wet

ISO 105-X12, grade 4 dry and grade 3-4 wet, evaluated against the ISO grey scale

Perspiration

ISO 105-E04, grade 4 for any colour likely to contact skin on handles

Colour change assessment

ISO 105-A02 grey scale, holding part test .one standard under the comparison workflow

Restricted substances

REACH Annex XVII covers azo dyes and chromium VI; Prop 65 covers lead in coated trim

Two particular colour families deserve special mention. Bright reds on leather are historically the weakest against light and against rubbing, so pushed-red briefs should specify grade 3-4 wet as an agreed exception rather than discovering it at test. And very pale linings are weak against transfer, which is why light-coloured interiors need an ISO 105-X12 wet result recorded even when the exterior colour is well behaved. Where hides are bought through tanneries audited by the Leather Working Group, their protocols cover part of this; ask for the reference either way.

A Colour Approval Sequence That Survives Contact With Bulk

Putting it together, the sequence below adds roughly four working days to a development calendar and removes the most expensive category of rework there is. Each step has an output that can be filed, and the file is what makes the second order simple.

1. Name the reference. A Pantone TCX number written onto the tech pack, agreed before any dye is mixed.
2. Approve a physical dip, minimum 10 x 10 cm, under D65 with a TL84 secondary check, within two working days of arrival rather than two weeks.
3. Record the reading. Delta E against the reference plus the metamerism index, written on the approval sheet next to the physical swatch.
4. Split and retain. One half of the signed swatch stays with the brand, one half with the programme file, and both are dated.
5. Write the tolerance onto the purchase order with the light source and formula, and list colour as a named defect class for inspection.
6. Check at pre-shipment against the retained swatch, at AQL 2.5 under ISO 2859-1, with the report issued to the brand by name.

Notice that four of the six steps are documentation. That is not bureaucracy: colour is the one specification where the evidence disappears fastest, because dye lots get used up, operators move on, and six months later nobody can reconstruct what was approved. A retained swatch solves that permanently.

When to Accept a Colour and When to Reject One

Judgement is still required, because a tolerance is a threshold and not a description of what matters. Accept where the reading sits inside tolerance, where fastness meets its grade, and where the difference is in depth rather than in hue — a slightly deeper tan reads as variation, whereas a shift toward green reads as a mistake, even at the same Delta E. Reject where the shift is in hue, where the metamerism index exceeds 1.0, or where two components that must read as one colour have drifted apart from each other, since buyers compare parts within the bag far more than they compare the bag to a chip.

There is a commercial dimension too. Holding a bulk shipment for a 1.2 Delta E deviation costs more than the shipment is worth once the selling window is taken into account, and the correct remedy at that point is usually a credit on the repeat rather than a re-run. This is why the tolerance belongs in writing early: it converts an emotional decision taken under time pressure into an arithmetic one taken at leisure. Every programme we run through our validated partner factory Quanzhou Yingming Bags files the tolerance and the swatch halves before the deposit clears, precisely so nobody has to invent a standard on the day the goods are packed.

Frequently Asked Questions

People Also Ask