Colour in Cosmetics and the Problem of Skin Tones

The hardest colour matching problem in consumer goods

Almost every colour reproduction task in industry involves matching a manufactured surface to a manufactured standard. Cosmetics face something considerably harder: matching a manufactured film to a living, translucent, variable surface that changes with the season, the time of day, the amount of oil on it and the light it happens to be standing in. Foundation is the extreme case — a product whose entire purpose is to be an invisible match to something that has no fixed value. That difficulty is real and technical. It is also, historically, entangled with a commercial failure that left large numbers of people unable to buy a product that worked for them, and the two threads have to be discussed together to make sense of either.

What a foundation actually is

Underneath the marketing, a complexion product is a straightforward pigment dispersion in a vehicle designed to spread thinly and stay put. The colour system is small and has been stable for decades.

  • Titanium dioxide. The opacifier and whitener, responsible for coverage. It is also the source of the flashback effect, because it scatters strongly in the ultraviolet and short-wavelength region that camera flashes deliver.
  • Iron oxides. Three of them — red, yellow and black — do nearly all the actual colour work, mixed in varying ratio to produce the full range of skin tones.
  • The vehicle. An emulsion, anhydrous base or powder matrix that carries the pigment, controls how it spreads, and determines the finish from matt to luminous.
  • Modifiers. Fillers, silicas, spherical powders and interference pigments that adjust the optical behaviour of the film without changing its underlying colour much.

That a four-pigment palette must cover all human skin explains a great deal. The system is capable of the range, but the ratios are delicate, and the difference between a shade that disappears and a shade that reads as a mask can be a small shift in the red-to-yellow balance.

Depth is the easy axis; undertone is the hard one

Shade ranges are usually organised first by depth — light to deep — because it is the obvious variable and the easy one to extend. The variable that determines whether a shade actually works is undertone: the hue bias underlying the skin, conventionally divided into cool or pink, warm or golden, and neutral, with olive frequently treated as a fourth category because it carries a green component that the standard three-oxide mix struggles to reproduce.

Undertone matters more than depth because the human visual system is extraordinarily sensitive to skin hue. A foundation half a step too light reads as a slightly pale face. A foundation the correct depth but the wrong undertone reads as grey, orange or ashy, and everyone notices without necessarily being able to say why. This is also why a range that offers many depths but only one or two undertone families fails for a large part of its intended market: the correct depth is available and still unwearable.

Why matching skin is technically unlike matching a panel

Skin is not a flat opaque reflector. Light entering it scatters within the dermis and re-emerges having travelled some distance sideways — subsurface scattering — which is why skin has a characteristic softness and why an opaque flat film sitting on top of it can look wrong even at a perfect surface colour match. A foundation that matches on a spectrophotometer can still read as a layer because it has removed the translucency.

  • Skin is not one colour. Forehead, jaw, neck and chest commonly differ by more than the gap between adjacent shades in a range, so “matching” means choosing which region to match.
  • It moves. Seasonal sun exposure, flushing, hormonal change and ageing all shift the target, sometimes by several shade steps across a year.
  • Formulas oxidise. Some bases darken or turn orange over hours on skin as ingredients react with air and sebum, so the correct shade in the shop is the wrong one by evening.
  • Application changes colour. Sheer application lets the underlying skin show through and shifts the result; full coverage delivers the pigment colour and nothing else.
  • Cameras disagree with mirrors. Flash photography, front-facing phone cameras and video lighting each render complexion products differently from the eye.

The historical shade deficit

For most of the modern cosmetics industry, mass-market complexion ranges were built around a narrow band of light to medium skin tones, with a small number of deeper shades appended if any. The practical consequence was that people with dark skin, and particularly people with dark skin and a cool or olive undertone, either bought from a handful of specialist brands, mixed their own from two products, or went without. This is not a contested historical claim; it is visible in the shade counts of the period.

How a commercial loop became a product gap

The causes were largely commercial rather than technical. Complexion products are inventory-heavy: every shade is a separate stock-keeping unit that has to be manufactured, packaged, distributed and stocked in every store, and shades that sell slowly tie up capital and eventually get discontinued. Retail buyers allocated limited shelf space to the fastest-moving shades, so deeper shades were stocked thinly, sold slowly because customers could not find them, and were then cut for poor performance. The result was a self-confirming loop in which weak sales caused by poor distribution were read as evidence of weak demand.

Where deeper shades did exist, they were frequently wrong in a specific way: extended in depth by adding black iron oxide to an existing warm base, producing shades that went grey or ashy rather than following the actual hue path of deeper skin, which generally requires more red and yellow rather than simply more black. A range could therefore claim a wide depth span and still be unusable, which is why depth counts alone are a poor measure of a range’s adequacy.

What changed and why it stuck

The industry shifted noticeably from around 2017, when a widely publicised launch opened with forty shades spanning the full depth range from the outset, treated as a baseline rather than an extension. Commercially it demonstrated that the demand had been there all along and simply unserved; competitively it made a twelve-shade launch look negligent. Within a few years, large ranges with deliberate undertone coverage had become the expectation across the mass and prestige markets alike.

  • Undertone families as structure. Better ranges are now built as a grid of depth against undertone rather than a single line, so a customer can move sideways as well as up and down.
  • Shade naming and ordering. Systematic codes that encode depth and undertone let customers navigate a large range; arbitrary names make a forty-shade wall unusable.
  • Adjacent categories caught up. Concealers, powders and colour correctors were extended too, because a matching foundation with no matching concealer solves half the problem.
  • The economics were solved by data. Better demand forecasting and direct-to-consumer sales reduced the inventory risk that had justified narrow ranges.

Measuring skin colour properly

Serious shade development now uses instruments rather than eyes alone. The dominant metric is the individual typology angle, or ITA, calculated from CIELAB lightness and the blue-yellow axis to place a skin measurement on a continuous scale from very light to dark. Unlike the Fitzpatrick scale — which was devised to predict sunburn response to ultraviolet treatment and describes reactivity, not colour, though it is constantly misused as a colour classification — ITA is an actual colorimetric quantity and can be used to check whether a range has gaps.

Development teams measure large, deliberately diverse panels of real skin, plot the population in colour space, and design shade grids to cover it with even spacing. This exposes two things that intuition misses: that the population is continuous rather than clustered into the categories marketing uses, and that gaps in a range are usually not at the extremes but in the deep-with-cool-or-olive-undertone region, which historically had the fewest products.

The retail problem nobody has solved

A technically excellent forty-shade range still fails at the point of purchase if the customer cannot identify their shade. Store lighting is frequently poor and warm, which flatters and misleads. Testers on the back of the hand match the hand, not the face. Online, the customer is judging shade swatches rendered on an uncalibrated screen, photographed under studio lighting on a model whose skin is not theirs, which is why complexion products carry some of the highest return rates in e-commerce.

  • Match on the jaw, in daylight. The only reliable in-store method is a stripe along the jawline checked at a window, because it bridges face and neck.
  • Instrument-assisted matching. Handheld colorimeters and calibrated in-app camera matching genuinely help, provided the calibration reference is used correctly.
  • Publish the numbers. Brands that state each shade’s depth and undertone systematically make remote matching far more tractable than swatch photographs do.
  • Swatches on varied skin. Showing every shade on skin from across the range is more informative than showing the range as smears on an arm.

Beyond foundation

Complexion is the visible battleground, but the same considerations run through the rest of the category. Blush and bronzer formulated at pigment loads suited to light skin simply do not show on deeper skin, so the entire product has to be reformulated rather than merely renamed. Lip colours behave differently over pigmented lips, where the underlying tone shifts a nominally neutral shade. Colour correctors work on complementary logic — peach and orange tones to neutralise blue-grey undereye discolouration on deeper skin, where a light pink corrector does nothing useful.

The technical frontier is now split between adaptive formulations, which use interference pigments or reactive systems to shift within a small range on application, and bespoke manufacture, where an in-store or online measurement drives a dispensing machine that mixes a single unit to a measured target. Adaptive products remain limited in range and are best understood as forgiving rather than universal. Bespoke dispensing solves the matching problem cleanly and replaces it with a supply chain problem, since a made-to-order product is difficult to reorder consistently and impossible to stock. Both approaches are attempts to escape the same underlying constraint: a continuous, moving target being served by a discrete set of physical products on a shelf.