RGB is light: screens add red, green and blue; the web defaults to sRGB.^7,8
CMYK is ink: print subtracts from paper, set for a named press such as FOGRA39.⁹
Perceptual models: OKLCH and CIELAB space colours as they look, for even steps.^7,11

RGB adds lightScreens. All three make white.

CMY subtracts inkPrint. All three make near-black, so presses add K.

RYB mixes paintThe art-class wheel. A tradition, not a standard.

Three models, three jobs. A brand file needs the first two.

The detail

RYB is paint: the red-yellow-blue wheel is a painters’ tradition, not a screen or press standard.

Colour theory mixes science and convention. The science is light, the three cone types in the eye and colorimetry. The wheels and harmonies are conventions: useful tools, not laws.

Screens are additive. Red, green and blue light add up to white, and the web assumes the sRGB space unless a colour says otherwise.^7,8 Print is subtractive: cyan, magenta and yellow inks absorb light, and black (K) is added because the three together only reach a muddy dark.

CIELAB was created because the older XYZ space is not perceptually uniform.⁷ Oklab, introduced in 2020, is now in CSS and Photoshop’s default gradients.¹¹ That is the space we build palette steps in.

A common myth says Newton made the colour wheel in 1666. His prism paper was published in 1672, and the seven-part circle appears in Opticks, prefaced 1704.³⁶

Hue, saturation, lightness: the three dials

Every colour is a hue, a strength and a lightness. Lightness does the most work, and HSL measures it badly.

The HSL trap: yellow and blue both read 50%, yet look near-white and near-black.⁷
Lightness: how close to white or black: the dial that makes text readable.¹
Saturation: how strong it is; in four studies it amplified brand personality.¹⁸

YellowHSL lightness 50%

In greyscaleluminance 0.93

BlueHSL lightness 50%

In greyscaleluminance 0.07

HSL calls both 50%. Your eye, and the contrast formula, see near-white and near-black.

The detail

Hue: where the colour sits on the wheel, from red round to violet.
Warm and cool: a convention for temperature contrast; useful in design, weak as psychology.

CSS Color 4 says HSL lightness “does not correspond to the visual perception of lightness across hues”. Pure yellow and pure blue both sit at 50%.⁷ Their relative luminance is 0.9278 and 0.0722.³⁷

Perceptual spaces fix this. In OKLCH, one just-noticeable difference is about 0.02, and hue holds steady as chroma drops.⁷ In the older LCH, a saturated blue turns noticeably purple as it fades.

Labrecque and Milne ran four studies with fictitious logos. Hue shaped the brand personality people saw; saturation and value amplified those traits.¹⁸

Warm and cool are relative. A blue-green reads cool next to orange and warm next to violet. We use temperature for contrast in a palette, not as a promise about mood.

Colour harmonies: where a palette starts

Harmonies are geometric rules for picking hues. They choose the hues; they do not make the result readable.

The catch: pure red on pure green, a classic complementary pair, scores 2.91:1.³⁷
Complementary: opposite hues: the strongest tension, best kept for one accent.
Analogous: neighbours on the wheel: calm, but they need lightness contrast.

Split-complementary The opposite’s two neighbours: tension, less clash.

Base hue

Aa#2163DE5.38:1 AA
Aa#DE3D214.38:1 Large only
Aa#C2DE211.53:1 Fail

White text on each swatch, ratio computed live. Same harmony, very different readability.

The detail

Split-complementary: a base plus its opposite’s two neighbours: tension with less clash.
Triadic and square: three or four evenly spaced hues: lively, and hard to keep on brand.

Some guides call harmonies formulas that ensure success. They are starting points for hue, nothing more. Readability comes from lightness, which the wheel does not show.

Monochromatic palettes use one hue at several lightnesses. They are the easiest to keep readable and the hardest to make distinctive.

The wheel above uses HSL hues. For final palette steps we move to OKLCH, so a set of tints looks evenly spaced rather than lumpy.⁷

Contrast decides who can read it

Body text needs 4.5:1 against its background. It is a measured ratio, and it fails at the second decimal.

Body text: at least 4.5:1 for WCAG 2.2 AA; 7:1 for AAA.^1,3
No rounding: 4.499:1 fails, so #777777 on white, at 4.48:1, fails.^1,37
Large text and UI: 3:1 for large text, icons, borders and focus rings.^1,2

Get seenLightest grey that passes#767676 on #FFFFFF4.54:1AA
Get seenOne step lighter: fails body text#777777 on #FFFFFF4.48:1Large only
Get seenPure red on white#FF0000 on #FFFFFF4.00:1Large only
Get seenRed on green, “complementary”#FF0000 on #00FF002.91:1Fail
Get seenPure blue on white#0000FF on #FFFFFF8.59:1AAA

Test your own pair

Text Background Get seen

Ratios from the WCAG 2.2 formula, cut to two decimals: thresholds are never rounded.

The detail

Logos are exempt: a logotype has no contrast rule, unless it also works as a link or button.^1,2
WCAG 3 is a draft: its contrast method is undecided, so WCAG 2.2 AA stays the bar.⁶

WCAG 2.2 contrast minimums^1,2,3
What it applies to	AA, the minimum	AAA, enhanced
Body text	4.5:1	7:1
Large text, 24 px or 18.5 px bold	3:1	4.5:1
Icons, borders, focus rings	3:1	Not set
Logos and brand names	Exempt	Exempt

The ratio compares relative luminance: (L1 + 0.05) / (L2 + 0.05). It runs from 1:1 for identical colours to 21:1 for black on white.³ Luminance weights green most: 0.2126 R + 0.7152 G + 0.0722 B, on linearised sRGB.⁵

Why 4.5? The W3C starts from 3:1 and adds 1.5 for the contrast sensitivity lost at 20/40 vision.¹ Large means 18 pt, or 14 pt bold, roughly 24 px or 18.5 px bold.

Colour must never be the only signal, a Level A rule. Links set apart by colour alone need 3:1 against the surrounding text, or an underline.⁴

APCA calls itself the candidate method for WCAG 3. The WCAG 3 draft of 10 September 2026 says its contrast algorithm is “yet to be determined”.⁶ Meet WCAG 2.2 AA today.

Our own palette, checked in the open

We hold our own colours to the same test. Two failed as text on navy, so we added lighter partners.

Lime on navy: our action pair scores 16.97:1, far past the 7:1 AAA line.³⁷
Raw blue on navy: 3.77:1 fails as text, so a lifted blue at 7.38:1 takes over.³⁷
Lime on cream: 1.07:1: lime never carries text on a light ground here.³⁷

Lime on navy16.97:1AAA
Navy on cream18.17:1AAA
Raw blue on navy3.77:1Large only
Lifted blue on navy7.38:1AAA
Raw grey on navy3.81:1Large only
Light grey on navy7.54:1AAA
Lime on cream1.07:1Fail
Blue on lime4.49:1Large only

The text-and-ground pairs this site uses, with the two raw colours that failed beside the partners that replaced them.

The detail

Grey on navy: 3.81:1 fails, so a lighter grey at 7.54:1 carries the small print.³⁷
Blue on lime: 4.49:1 misses by a hair, so that pair never carries body text.³⁷

Every pair above is computed from the hex values in our stylesheet, the same file that colours this page.³⁷ The grid is the check we run on any palette before it reaches a page.

Raw blue on cream scores 4.81:1. It passes, but tight for small type, so small text uses a darker blue ink at 6.95:1.

None of this changes the brand. Navy, blue, lime, cream and grey stay the five colours people see. The derived shades are tools that let each one do its job readably.

Our colour palette lesson walks through the same idea step by step.

Colour blindness: test the palette, not the eye

About 1 in 12 men see colour differently. Simulate the palette and make sure nothing depends on hue alone.

Never colour alone: a WCAG Level A rule: pair it with a word, shape or icon.⁴
Who: about 1 in 12 men and 1 in 200 women.^12,14
Why ours holds: lime and navy differ in lightness, so every simulation keeps them apart.^16,37

Typical vision

OverduePaid

Protanopiano red cones

OverduePaid

Deuteranopiano green cones

OverduePaid

Tritanopiano blue cones

OverduePaid

Try a colour

Simulated with the Machado 2009 model at full strength. Navy and lime stay apart; the dots need their words.

The detail

Most common: deuteranomaly, a weakened response to green.¹³

Red-green deficiency is inherited on the X chromosome, which is why men are affected far more. Blue-yellow deficiency affects men and women equally.¹⁴

A large survey review found about 8% of European men and 0.4% of women affected, and 4–6.5% of Chinese and Japanese men.¹⁵ Worldwide estimates range from 200 million to 300 million people.^14,16

The simulation uses the Machado, Oliveira and Fernandes model at full severity, applied in linear sRGB.¹⁶ It shows the most severe case; milder forms sit between typical vision and these rows.

The red and green dots are an example pair, not ours. Under deuteranopia they read as near twins, which is why status colours always need a word or an icon.

Colour psychology: what the evidence supports

Colours carry associations, but product, positioning and culture change them. Several famous colour statistics have no study behind them.

Shared, not fixed: colour–emotion links agree across 30 nations, yet nation still shifts them.²²
Climate counts: yellow meant joy to about 6% in Egypt and 88% in Finland.²³
Fit beats formula: colours that match a brand’s positioning look more appropriate.²⁰

The famous claims, checked

“62–90% of a first judgement is colour”¹⁷Review, no data
“85% of shoppers buy because of colour”³³No study found
“Red lowers test scores”²⁷Did not hold up
Colour should fit the brand’s position²⁰Two experiments
Colour–emotion links are shared²²Shown in 30 nations

Popular charts list meanings. Studies say product, positioning and culture change them.

The detail

The 62–90% claim: comes from a literature review, with no experiment behind the figure.¹⁷
The 85% claim: we found no study behind “85% of shoppers buy on colour”.³³

What popular colour charts say. Associations, not rules.³³

Redenergy, urgency, love
Orangewarmth, play
Yellowoptimism, caution
Greennature, honesty
Bluetrust, calm
Purpleluxury, creativity
Blackpremium, power
Whitesimplicity, clean

Jonauskaite and colleagues asked 4,598 people in 30 nations to link 20 emotions with 12 colour terms. Associations were broadly universal, with an average similarity of r = .88, yet nation still predicted them beyond the shared pattern.²²

A second study across 55 countries, as reported, found yellow linked to joy far more often in rainy, northern places.²³ Treat those two percentages as a secondary report.

Bottomley and Doyle ran two experiments. Functional products suited functional colours, and people judged colours matching a brand’s position as more appropriate.²⁰ An eight-country study found both similarity and difference in colour meaning.²¹

Red seemed to impair test scores in four experiments in 2007.²⁶ A 2020 meta-analysis found no evidential value after correcting for publication bias.²⁷ A review calls the field “nascent”.²⁵

The 62–90% figure traces to a 2006 literature review that cites no study for it.¹⁷ The 85% figure is credited to various blogs; we found no original.³³

How many colours a brand should have

Give one colour the job of recognition, then add neutrals and an accent. Roles matter more than the count.

One primary: the colour people should remember you by.
One accent: saved for the action you want, so it stays loud.
60-30-10: a starting heuristic; we found no study behind it.³⁴

60%Cream30%Navy10%Lime

Our palette at 60-30-10. Swap the proportions and the same colours shout.

The detail

Neutrals: carry text and grounds; most of any page is neutral.
Functional colours: success, warning, error and focus are tools, not brand colours.

Shopify calls 60-30-10 a design guideline: 60% primary, 30% secondary, 10% accent. It gives no source.³⁴ We could trace it only to interior-design blogs, with no author, date or test.

It still works as a first draft. The useful idea is that one colour dominates, one supports, and one is rare enough to point.

Roles first: a primary for recognition, a support colour for links and detail, an accent for the one action, neutrals for grounds and fine print. Functional colours for success, error and focus are set per site.

On our own site, navy and cream share the ground, blue supports, and lime is kept for the one button that matters.

Screen and print need different numbers

Specify every brand colour once per medium: HEX for screens, CMYK for a named press, Pantone where ink must match.

CMYK depends on the press: name the condition, such as FOGRA39 on coated paper.⁹
Wider screens: Display P3 holds about 1.5 times the colours of sRGB.⁷
Proof it: check printed colour on the real stock before a full run.

HagoDigital blue

ScreenHEX and RGB, in sRGB: #2365DE35 101 222
PrintCMYK, for a named press: From a proof, never a converter guess
Spot inkPantone, coated and uncoated: Matched on a printed guide

How we hand over one colour: one value per medium, print values only from a proof.

The detail

HEX is sRGB: the web’s default colour space, standardised by the IEC in 1999.^7,8
Pantone spot inks: 2,614 premixed colours, sold on coated and uncoated guides.¹⁰

CSS Color 4 gives Display P3 a gamut of 1.233 million CIELAB units against 0.820 million for sRGB, about 1.5 times.⁷ Both share the D65 white.

FOGRA39 describes offset printing on coated paper of 115 g/m².⁹ A CMYK value only means something next to the condition it was made for.

Pantone is not perfectly exact either. Its guides come in coated and uncoated versions because the paper changes how one ink looks.¹⁰

We leave the CMYK and Pantone slots empty until there is a proof. A converted number nobody has printed is a guess.

Can a brand legally own a colour?

Sometimes. A colour can be a trademark when people link it to one brand and it does no functional job.

Qualitex, 1995: the US Supreme Court allowed a green-gold colour on press pads.²⁸
Cadbury purple: lost in 2013: “predominant” made the description too imprecise.²⁹
Tiffany Blue: registered since 1998; its Pantone 1837 Blue is not public.³⁰

Telekom magenta³¹#E20074 on white4.68:1AA
Telekom magenta³¹#E20074 on black4.49:1Large only
Telekom dark mode³¹#F61488 on black5.37:1AA
IBM Blue 60³²#0F62FE white text on it5.00:1AA

Only values the owners publish. Telekom’s magenta misses AA on black, so its dark mode uses a lighter one. Cases explained here, not legal advice.

The detail

Telekom magenta: published as a value, and not free for others to use in commerce.³¹
Precision wins: a protectable colour is defined exactly: which shade, where, how.²⁹

In Qualitex, decided 28 March 1995, the Court held that a colour can meet trademark rules if it has acquired meaning and is not functional.²⁸ It cited Owens-Corning’s pink insulation from 1985.

Cadbury applied in 2004 for purple, Pantone 2685C, on the whole or “predominant” surface of packaging. The Court of Appeal held that too subjective to be a sign.²⁹

Telekom’s magenta passes on white at 4.68:1 but fails on black at 4.49:1. Its dark theme uses a lighter magenta that scores 5.37:1.^31,37

Only Telekom and IBM publish a screen value, so only they appear above. Tiffany keeps its Pantone 1837 Blue private, Cadbury filed a print ink, Pantone 2685C, and we found no published value for Qualitex’s green-gold.^28,29,30

This explains public cases; it is not legal advice.

How to choose brand colours, step by step

Four phases, nine steps: we start from meaning and finish with measurements every colour must pass.

Meaning first: write the positioning before anyone opens a colour picker.²⁰
Read the category: list rivals’ colours, then choose to fit in or stand apart.¹⁹
Test, then specify: contrast, a colour-blind simulation, then HEX, CMYK and Pantone.^1,16

Steps 1–2PositionHow it should feel, and against whom.
Steps 3–5PickOne primary, tints in OKLCH, then neutrals.
Steps 6–7ProveEvery text pair 4.5:1; simulate colour blindness.
Steps 8–9Hand overHEX, CMYK and Pantone, written into the brand book.

How we take a palette from meaning to file, ours included.

The detail

Build in OKLCH: tints and shades with even lightness steps.⁷

Positioning first (Position): how should it feel, and to whom?
Read the category (Position): list rivals’ colours; fit in or stand apart.
One primary hue (Pick): check it carries no unwanted meaning where you sell.
Tints and shades (Pick): built in OKLCH, so lightness steps look even.
Neutrals and functions (Pick): text, grounds, success, error, focus.
Contrast (Prove): every text pair 4.5:1; large text and UI 3:1.
Colour blindness (Prove): simulate it; never colour alone.
Every medium (Hand over): HEX, CMYK for a named press, Pantone.
Write it down (Hand over): roles, pairings and proportions in the brand book.

Labrecque and Milne studied the colours of 281 top brands across 15 categories. Standing apart helped in some categories and hurt in others, especially where a dominant leader sets the norm.¹⁹

Step three includes a culture check. A hue that reads well at home can carry a different meaning in another market.^21,22

Step nine matters if you ever want to protect a colour. Cadbury lost because its description was vague.²⁹

Juan Pablo Tagliaferro, arms folded, in a black t-shirt against a dark ground.

What I’d do

I’d write one sentence about how the brand should feel before looking at a single swatch.
I’d pick one colour to be remembered by, and make everything else serve it.
I’d test every text pair before anyone falls for a mock-up.
I’d ignore any colour statistic that can’t name its study.

Juan Pablo Tagliaferro, founder, HagoDigital

Questions people ask next

How many colours should a brand have?

One primary for recognition, one or two neutrals for text and grounds, and one accent for action. Add functional colours for success, warning and error. Roles matter more than the number.

What is the 60-30-10 rule?

A proportion heuristic: 60% dominant, 30% secondary, 10% accent.³⁴ We found no traced origin or study testing it, so use it as a first draft and adjust by eye.

Does colour psychology actually work?

Colour carries meaning, in context. Associations shift with product, positioning and culture, and some famous lab effects did not replicate.^20,27 Use meaning as a direction, not a formula.

Is a complementary palette good for a brand?

It can be, as a base and an accent. Opposite hues do not make text readable on their own: pure red on pure green scores 2.91:1.³⁷ Lightness contrast decides readability.

Which colour wheel should I use?

For screens, RGB in sRGB. For print, CMYK for a named press condition. For even palette steps, OKLCH. The red-yellow-blue wheel is a painting tradition.

Sources

Standards: W3C, IEC, ICC, Pantone

W3C WAI, Understanding SC 1.4.3 Contrast (Minimum): text needs at least 4.5:1; large-scale text at least 3:1; large = 18 pt, or 14 pt bold (1pt = 1.333px, about 24 px or 18.5 px bold); 4.5 = 3 × 1.5, the contrast sensitivity loss at 20/40 vision; thresholds are not rounded, so 4.499:1 fails; text in a logo or brand name has no contrast requirement. Link (read 30/10/2026)
W3C WAI, Understanding SC 1.4.11 Non-text Contrast: UI components and meaningful graphics need 3:1 against adjacent colours; 2.999:1 fails; a logo used as a control is not exempt. Link (read 30/10/2026)
W3C, Web Content Accessibility Guidelines 2.2, Recommendation 12 December 2024: SC 1.4.6 Contrast (Enhanced) 7:1, large text 4.5:1; contrast ratio (L1 + 0.05) / (L2 + 0.05), from 1:1 to 21:1. Link (read 30/10/2026)
W3C WAI, Understanding SC 1.4.1 Use of Color, Level A: colour is not the only visual means of conveying information; links set apart by colour alone need 3:1 against surrounding text (G183). Link (read 30/10/2026)
W3C WAI GL wiki, Relative luminance: L = 0.2126 R + 0.7152 G + 0.0722 B on linearised sRGB; threshold 0.04045. Link (read 30/10/2026)
W3C, WCAG 3.0 Working Draft, 10 September 2026: “The contrast algorithm used in WCAG 3 is yet to be determined”. APCA describes itself as the candidate method (git.apcacontrast.com). Link (read 30/10/2026)
W3C, CSS Color Module Level 4, Candidate Recommendation Draft 29 Oct 2026: untagged colours are sRGB; primaries (CIE 1931 xy chromaticity) sRGB red 0.640, 0.330, green 0.300, 0.600, blue 0.150, 0.060; Display P3 red 0.680, 0.320, green 0.265, 0.690, blue 0.150, 0.060; gamut 1.233 million vs 0.820 million CIELAB units (about 1.5 times, our arithmetic); HSL lightness does not match perceived lightness, yellow and blue both 50%; OkLCh improves hue linearity, one JND about 0.02; CIELAB built because XYZ is not perceptually uniform (CIE S 014-4). Link (read 30/10/2026)
IEC 61966-2-1, Default RGB colour space – sRGB, published 1999-10-18. Link (read 30/10/2026)
International Color Consortium registry, FOGRA39: offset printing per ISO 12647-2, coated paper, 115 g/m², 60/cm screen. Link (read 30/10/2026)
Pantone, Formula Guide coated and uncoated: 2,614 spot colours on coated and uncoated stock. Link (read 30/10/2026)
Björn Ottosson, A perceptual color space for image processing (Oklab, 2020): now in CSS Color 4 and 5 and Photoshop gradients. Link (read 30/10/2026)

Colour vision

National Eye Institute (NIH), Color blindness: about 1 in 12 men have colour vision deficiency; the most common type confuses red and green. Updated 5 Nov 2025. Link (read 30/10/2026)
National Eye Institute, Types of color vision deficiency: deuteranomaly is the most common red-green type. Link (read 30/10/2026)
Colour Blind Awareness (UK charity): approximately 1 in 12 men (8%) and 1 in 200 women; about 300 million worldwide; red-green genes on the X chromosome, blue-yellow affects men and women equally. Link (read 30/10/2026)
Birch J (2012), Worldwide prevalence of red-green color deficiency, JOSA A 29(3): European men about 8%, women about 0.4%; Chinese and Japanese men 4–6.5%. Link (read 30/10/2026)
Machado, Oliveira and Fernandes (2009), A physiologically-based model for simulation of color vision deficiency, IEEE TVCG 15(6): about 200 million people affected; severity 1.0 matrices used here, applied in linear sRGB. Link (read 30/10/2026)

Colour psychology: peer-reviewed

Singh S (2006), Impact of color on marketing, Management Decision 44(6): a literature review; abstract states people decide within 90 seconds and 62–90% of the assessment is based on colour, citing no study. Link (read 30/10/2026)
Labrecque and Milne (2012), Exciting red and competent blue, J. Academy of Marketing Science 40: four studies; hue shapes brand personality, saturation and value amplify traits. Link (read 30/10/2026)
Labrecque and Milne (2013), To be or not to be different, Marketing Letters 24(2): 281 top brands, 15 categories and 40 subcategories; differentiation helps some categories and harms others. Link (read 30/10/2026)
Bottomley and Doyle (2006), The interactive effects of colors and products on perceptions of brand logo appropriateness, Marketing Theory 6(1): two experiments; colours congruent with a brand’s positioning are judged more appropriate. Abstract via Crossref. Link (read 30/10/2026)
Madden, Hewett and Roth (2000), Managing images in different cultures, J. International Marketing 8(4): eight countries; similarity and dissimilarity in colour preference and meaning. Abstract via Crossref. Link (read 30/10/2026)
Jonauskaite et al. (2020), Universal patterns in color-emotion associations, Psychological Science 31(10): 4,598 participants, 30 nations, 22 languages, 20 emotions × 12 colour terms; average similarity r = .88, nation still predicts associations. Link (read 30/10/2026)
Jonauskaite et al. (2019), J. Environmental Psychology 66, as reported by L. T. White, Psychology Today, 12 Jan 2021 (secondary report): 6,625 participants in 55 countries; yellow linked with joy by about 6% in Egypt and about 88% in Finland. Link (read 30/10/2026)
Palmer and Schloss (2010), An ecological valence theory of human color preference, PNAS 107(19): people like colours of liked things; 48 participants, 32 colours, r = +0.893. Link (read 30/10/2026)
Elliot and Maier (2014), Color psychology, Annual Review of Psychology 65: the literature “remains at a nascent stage”. Link (read 30/10/2026)
Elliot et al. (2007), Color and psychological functioning, JEP: General 136(1): four experiments; red before a test impaired performance. Link (read 30/10/2026)
Gnambs (2020), Limited evidence for the effect of red color on cognitive performance, Psychonomic Bulletin & Review 27(6): meta-analysis of 67 effect sizes; no evidential value after correcting for publication bias. Link (read 30/10/2026)

Courts and brands’ own published values

Qualitex Co. v. Jacobson Products Co., 514 U.S. 159, decided 28 March 1995: a colour can meet trademark requirements with secondary meaning and no function; cites Owens-Corning pink (1985). Link (read 30/10/2026)
Société des Produits Nestlé v Cadbury [2013] EWCA Civ 1174, 4 October 2013: application of 15 October 2004 for purple, Pantone 2685C, as the “predominant” colour of packaging, held too subjective and imprecise to be a sign. Link (read 30/10/2026)
Tiffany & Co. press site: Tiffany Blue registered as a colour trademark since 1998; standardised in 2001 as Pantone 1837 Blue, not publicly available. Link (read 30/10/2026)
Deutsche Telekom design tokens (npm @telekom/design-tokens): primary #e20074, dark theme primary #f61488; Scale README: magenta not available for free use in commerce. Link (read 30/10/2026)
IBM Carbon Design System, colour tokens: Blue 60 = #0f62fe, used for primary buttons and focus; focus states must pass 3:1. Link (read 30/10/2026)

Popular claims we checked

Selfpackaging, La teoría del color (Spanish, modified 05/05/2026): repeats 62–90% (credited to an “Institute for Color Research”) and “85% of shoppers” buy on colour (credited to Neil Patel/QuickSprout); associations listed: blue trust, black premium, green and brown honesty, yellow optimism. No primary study found for 85%. Link (read 30/10/2026)
Shopify, How to choose brand colors (21/02/2025): 60-30-10, primary 60%, secondary 30%, accent 10%, called a design guideline, no source given. Link (read 30/10/2026)
Interaction Design Foundation, Color theory: states “about 8% of people” are colour blind, mixing the rate for men with people. Link (read 30/10/2026)
Newton, New theory about light and colors, Phil. Trans. R. Soc. 6, published 19 Feb 1672; Opticks, preface dated 1 April 1704, divides a circle into seven parts. Often misdated to 1666. Link (read 30/10/2026)

Our own numbers

HagoDigital, our own stylesheet (site/styles.css): navy #000A18, blue #2365DE, lime #CEFF00, cream #F8F5E6, grey #696D71, blue ink #1B4EB0, blue lift #6E9BFF, cream mute #9AA0A8; HagoDigital blue #2365DE is RGB 35, 101, 222. Contrast ratios and luminance computed by us with the WCAG 2.2 formula (sources 3 and 5), cut to two decimals: luminance of #FFFF00: 0.9278 (0.93); luminance of #0000FF: 0.0722 (0.07); white on #2163DE: 5.38; white on #DE3D21: 4.38; white on #C2DE21: 1.53; #767676 on #FFFFFF: 4.54; #777777 on #FFFFFF: 4.48; #FF0000 on #FFFFFF: 4.00; #FF0000 on #00FF00: 2.91; #0000FF on #FFFFFF: 8.59; navy #000A18 on cream #F8F5E6: 18.17; navy #000A18 on white #FFFFFF: 19.87; navy #000A18 on blue #2365DE: 3.77; navy #000A18 on lime #CEFF00: 16.97; cream #F8F5E6 on navy #000A18: 18.17; cream #F8F5E6 on white #FFFFFF: 1.09; cream #F8F5E6 on blue #2365DE: 4.81; cream #F8F5E6 on lime #CEFF00: 1.07; lime #CEFF00 on navy #000A18: 16.97; lime #CEFF00 on cream #F8F5E6: 1.07; lime #CEFF00 on white #FFFFFF: 1.17; lime #CEFF00 on blue #2365DE: 4.49; blue #2365DE on navy #000A18: 3.77; blue #2365DE on cream #F8F5E6: 4.81; blue #2365DE on white #FFFFFF: 5.27; blue #2365DE on lime #CEFF00: 4.49; blue ink #1B4EB0 on navy #000A18: 2.62; blue ink #1B4EB0 on cream #F8F5E6: 6.95; blue ink #1B4EB0 on white #FFFFFF: 7.60; blue ink #1B4EB0 on blue #2365DE: 1.44; blue ink #1B4EB0 on lime #CEFF00: 6.49; blue lift #6E9BFF on navy #000A18: 7.38; blue lift #6E9BFF on cream #F8F5E6: 2.46; blue lift #6E9BFF on white #FFFFFF: 2.69; blue lift #6E9BFF on blue #2365DE: 1.95; blue lift #6E9BFF on lime #CEFF00: 2.30; grey #696D71 on navy #000A18: 3.81; grey #696D71 on cream #F8F5E6: 4.77; grey #696D71 on white #FFFFFF: 5.22; grey #696D71 on blue #2365DE: 1.01; grey #696D71 on lime #CEFF00: 4.45; cream mute #9AA0A8 on navy #000A18: 7.54; lime on navy, simulated protan: 16.38; red #D7263D vs green #2BA84A, simulated protan: 2.49; lime on navy, simulated deutan: 16.58; red #D7263D vs green #2BA84A, simulated deutan: 1.27; lime on navy, simulated tritan: 16.50; red #D7263D vs green #2BA84A, simulated tritan: 1.45; HagoDigital blue as RGB: 35, 101, 222; #E20074 on #FFFFFF: 4.68; #E20074 on #000000: 4.49; #F61488 on #000000: 5.37; #FFFFFF on #0F62FE: 5.00. Link (computed 30/10/2026)

HagoDigital builds brands and their palettes, so we have an interest in this question; that is why every figure here is sourced.
Our prices live on the pricing page, not here. To see how we build a brand, this is what we do.

Get a quote