How foundation shade finders work, and how ours does
Every shade in our data has a colour. When you pick the shade you wear, we measure how far every other shade is from it. Then we show you the closest one in each range. Here's exactly what that means, and what it can't do.
Four ways shade finders work
Online shade finders mostly work in one of four ways, and each has a weak spot.
- What other people wear. Some finders record which shades people say match them in several brands. That captures real wear, but it only works for shades enough people have reported.
- A quiz. Questions about your skin lead to a recommendation, usually within one brand or one shop's stock.
- A photo or an in-store scanner. A camera or device reads your skin. Lighting and the camera change what it sees, so results vary from photo to photo.
- Colour data. Each shade gets a measured colour and the finder compares colours. That's how ours works. It can compare every shade in its data, and it can show how close each result is.
1. Each shade has a measured colour
Our data holds 6,950 shades from 109 brands. Most have a colour taken from the swatch image the retailer showed online in January 2021. For some products the colour comes from the brand's own website (October 2026), and each product page says which. We convert that colour to CIELAB, a way of describing colour that's built around how people see it. It uses one number for lightness and two for hue.
2. We measure the difference the way colour scientists do
To compare two shades we use CIEDE2000, the standard formula for how different two colours look to the human eye. A bigger number means a bigger difference. In every range we keep the shade with the smallest difference from yours, then rank the ranges from closest to furthest.
3. What the labels mean
Swatch photos vary. Lighting, camera and screen differ from one product page to the next. So even two colours that are meant to be the same don't come out identical. We measured how much. We compared 4,780 pairs where one brand uses the same shade code in two formulas. The two swatches differ by a median of 7 on the CIEDE2000 scale, and half the pairs fall between 4.2 and 10.4. Only 27% are under 4.5. So our labels are set from that measured spread, not from a fixed cut-off:
| Label | Difference | What to do |
|---|---|---|
| Likely close | under 4.5 | As close as our data can tell two shades apart. A good first shade to try. |
| Worth sampling | 4.5 to 10 | In the right area. Sample it next to the shade you wear before you buy. |
| Unlikely | 10 or more | Further apart than most shades that are meant to match. This range probably has nothing near yours. |
Some shade finders label a difference under 1 "indistinguishable". Photo-based colour data can't support that precision, so we don't claim it. A "likely close" result is as close as the photos can show, which is why we still ask you to test it.
4. Which way a match leans
Under each match we say what its swatch reads compared with yours: lighter or deeper, warmer (more golden) or cooler (more pink). Small differences that are within the noise of the photos are shown as "about the same".
5. Undertone comes from the listing first
Where a shade's name, code or retail listing gives its undertone (cool, neutral, warm or olive), we show it. Swatch photos are poor at showing undertone, so the listing comes first. If the swatch clearly leans the other way, we say so. For example, Estée Lauder Double Wear 2C0 Cool Vanilla is listed as cool, but its 2021 swatch reads warm.
What it can't do
- It can't promise a match. It narrows a wall of bottles to two or three worth trying.
- It doesn't know how a foundation wears, oxidises or looks once it sets on your skin.
- Most of our colours date from January 2021. Products checked on the brand's own website in October 2026 say so on their page. For the rest, newer launches and reformulations may be missing or have changed.
- Some 2021 swatch images didn't come out (they read as white or grey). We leave those shades out of the matching.
The best check is still a swatch on your jawline in daylight, left for an hour to see how it settles. Where our data comes from.