How cross-brand paint matching works

Every color page on this site lists the closest colors to it in each of the other brands. This is how those matches are calculated, what the wording on them means, and - the part most match tools leave out - where the method stops being reliable.

The measurement

Comparing two colors by their hex codes does not work. Hex is a set of screen instructions, and the numeric distance between two of them has no consistent relationship to how different they look: a gap of 10 in the red channel is obvious in a pale grey and invisible in a deep navy.

So every color here is converted to CIE L*a*b* (D65), a space built so that distances approximate perceived difference, and the pairs are compared with CIEDE2000 - the current CIE standard for color difference, which corrects Lab’s known unevenness around lightness, chroma and the blue region. The result is a single number, ΔE00. Zero means identical. Bigger means further apart.

Where a manufacturer publishes measured Lab values, those are used directly. None of the brands in the catalog currently do - they publish hex or RGB - so every Lab here is derived, and each color page says so. A derived value inherits whatever approximation the brand made when it produced its swatch file, which is the floor on how accurate any of this can be.

What the bands mean

The wording on each match maps to a ΔE00 range:

That last band matters. A match list that only ever showed close results would be lying by omission when a brand has nothing near your color; instead every brand gets its nearest colors with an honest label. Read the band, not the position in the list.

One asymmetry worth knowing: the same ΔE is more noticeable on a deep color than on a pale one. A ΔE of 1.5 between two off-whites is nothing. Between two deep navies, on a full wall, it is the difference between “the same color” and “why does that wall look wrong”. Treat the bands as slightly optimistic at the dark end.

How the matches are built

The matching runs once, when the site is built, not when you load the page. Every color is compared against every color in every other brand, and the closest three per brand are kept, sorted by difference. Nothing is computed live and there is no lookup to get out of date between deployments - the answer for a given catalog is fixed, so it is precomputed.

To keep that tractable, candidates more than 10 L* away in lightness are skipped: ΔE00 cannot be small when lightness is far apart, so those comparisons cannot produce a match. If a brand has nothing at all inside that window - which happens for very dark and very light colors, where a brand’s range thins out - the search widens to the full catalog for that brand, so no brand is ever silently missing from a list.

What ΔE cannot tell you

This is the part that decides whether to trust a match, and none of it is in the number.

Finish changes the color. The same pigment in flat, eggshell and gloss returns light differently, and reads as a different depth on a wall. A ΔE 0.8 match between a flat and a semi-gloss will not look like a match.

Metamerism. Two paints can be measured as near-identical under one light and visibly different under another, because they get there with different pigments. Nothing computed from a single set of values predicts this, and it is the most common reason a “perfect match” fails on site.

Base and colorant differ between brands. Two manufacturers mixing to the same target use different bases, tints and dispersions. The tinted result depends on the whole system, not on the target value.

The source data is a swatch file, not the wall. These values come from the files manufacturers publish for designers. They describe an idealized swatch. Real paint varies by batch, substrate, number of coats, and the color underneath it.

So when should you trust it

Use a match to shortlist: to find the two or three candidates at another brand worth buying samples of, or to work out whether an equivalent exists at all before driving to a store. That is what it is good for, and it is genuinely useful.

Do not use it as a substitute for a sample. Before you commit - especially for a repair, a touch-up, or matching an existing wall - get the physical chip or a sample pot, put it against the real thing, and look at it in the light the room actually has. If the two are for the same room and must not clash, that step is not optional.

Related: how to read LRV and warm vs cool undertones.