The recognition–reproduction gap
Ask someone what color Spotify uses and the answer arrives in under a second: green. Ask them to pick that green out of a gradient of forty greens and confidence collapses. This is not a failure of attention. It is how visual memory is built. The eye is extraordinarily good at comparing two colors that are present at the same time and comparatively poor at storing one for later retrieval.
What memory stores instead is a category. Colors get filed under a linguistic label — red, green, navy, turquoise — and the label is what survives. Everything inside a label collapses toward that category's prototype, usually a version more saturated and more typical than the original. That is why guesses in a color memory game cluster in a predictable direction rather than scattering randomly around the target.
What 259 brand colors actually look like
The brand pool behind LogoColorQuiz holds 259 colors across nine categories, each one read from publicly visible usage: websites, app icons, packaging, storefronts, and product interfaces. Sorted by hue, the distribution is far from even:
- Blue and cyan — 35 percent. By a wide margin the most common brand hue family. Blue carries associations of reliability and low risk, and it is the least likely color to be culturally objectionable in any large market.
- Red — 22 percent. The appetite and urgency color. Heavily concentrated in food, retail, and entertainment.
- Green — 10 percent, orange — 9 percent, purple — 5 percent, yellow — 4 percent, pink — 3 percent.
- Near-black, near-white, and grey — 12 percent. A category that has grown quickly, driven almost entirely by newer software and AI brands.
That concentration is exactly what makes the game hard. If a third of all brands are blue, knowing that a brand is blue narrows almost nothing. The work is in the remaining two axes.
Saturation is where categories separate
Average saturation per category tells a sharper story than hue does:
- Food and drink — 87 percent average saturation, average lightness 41 percent. Shelf and storefront colors are loud on purpose; they compete for attention at three meters.
- Finance — 84 percent saturation, and 39 percent of finance brands are blue, the highest single-hue concentration of any category.
- Technology — 81 percent saturation, split between blue at 29 percent and cyan at 24 percent.
- AI — 49 percent saturation, average lightness 34 percent, with 35 percent of the category landing in the neutral band. AI brands deliberately avoid the loud palette their industry could easily have adopted.
- Cartoon characters — 62 percent saturation but the highest average lightness at 54 percent. Character colors are bright rather than intense.
The practical consequence: category knowledge is a real advantage in a color quiz, but only if you use it on saturation and lightness rather than hue. Knowing a brand is a fast-food chain should push a guess toward extreme saturation. Knowing it is an AI lab should push it toward grey.
Why brands pick the colors they pick
Three forces do most of the work, and none of them is color psychology in the pop-science sense.
Distinctiveness inside a category
A brand's color is chosen against its competitors, not in the abstract. When a category converges on blue, the newcomer's cheapest differentiator is to not be blue. That is visible in the data: chip makers cluster in blue, and the two best-known challengers picked bright green and bright red instead.
Reproduction constraints
A brand color has to survive embroidery, single-color print, an app icon at 29 pixels, an illuminated sign at night, and a phone screen in sunlight. Colors that only work in one medium do not last. This is why so many brand colors sit in the mid-lightness band rather than at the extremes — extremes break first.
Legal availability
Color can be protected as trade dress in some jurisdictions when it has acquired distinctiveness in a specific category. In practice this narrows options for late entrants in categories where an incumbent's color is strongly established.
Measuring how wrong a guess is
Comparing colors by their hex values is misleading. The distance between #FF0000 and #EE0000 looks small in hex and small to the eye; the distance between #00FF00 and #00EE00 looks identical in hex but is far less visible, because human vision is much more sensitive to differences in some regions of color space than others.
The standard fix is CIE Lab, a color space built so that equal numeric distances correspond roughly to equal perceived differences. The distance between two colors in that space is called Delta E. As a rough reading scale: below 2.3 most people cannot see a difference at all; below 10 the two colors read as the same color; 20 to 30 reads as the right family but visibly off; above 50 the colors no longer look related.
Every guess in every mode on this site is converted from the slider values to RGB, linearised, converted to CIE Lab against a D65 white point, and compared to the stored target with a Delta E distance. The score you see is that distance mapped onto a 0–100 curve — which turned out to be the single hardest design decision in the whole game, because a linear mapping rated ordinary human performance as total failure.
What players actually get wrong
Aggregated gameplay data shows two consistent biases. Guesses skew too saturated, because memory reaches for the category prototype rather than the real color, and they skew too light, because screens display brand colors against white far more often than against dark backgrounds and the surrounding white pulls the remembered value upward.
Neither bias disappears with practice, but both become correctable once you know they exist. The reliable technique is to make your best guess, then deliberately pull saturation and lightness down before submitting. Most players improve more from that single adjustment than from any amount of additional play.
Try it
The fastest way to see your own bias is five rounds against real brand colors, with the target revealed beside your guess after each one.