Ana teaches yoga in Ljubljana. When she rebuilt her studio's website, she read three articles about color psychology. One said red buttons convert. One said blue builds trust. One said up to 90% of snap judgments about a product are based on color alone.
So she made the page blue for trust, the buttons red for conversion, and added a soft orange because a fourth article said orange feels friendly. The result looked like a pharmacy having a sale. Visitors could not tell what to click first, and the pale orange text on her class timetable was close to unreadable on a sunny phone screen.
Ana did not need more color psychology. She needed less of it, and more of three plain ideas: color meaning depends on context, attention goes to what is different, and text has to be readable before it can feel like anything.
Ana and her studio are imagined. The research, standards and court cases below are real.
Colors have no fixed meaning
The most influential academic account of color and behavior is the color-in-context theory of Andrew Elliot and Markus Maier, set out at length in a 2012 chapter. Its core claim is in the name: what a color means, and how people respond to it, depends on where they meet it. Red in an exam setting can signal danger and failure; red in a romantic setting can signal attraction.
The best-known evidence came from Elliot and colleagues in 2007. Across four experiments, a brief exposure to red before an important test, such as a red participant number on an IQ test, was linked to lower scores, and participants did not seem aware of the influence. Two further experiments linked red to avoidance behavior.
It is a striking finding. It has also not held up well. Timo Gnambs's 2020 meta-analysis pooled 67 effect sizes from 38 samples. For anagram and knowledge tests, red made no significant difference. For reasoning tests there was a modest negative effect, but early studies showed much larger effects than later ones, and after correcting for publication bias, the analysis found no evidential value for an effect of red on intellectual performance.
For a website, the useful part of the theory survives the replication trouble: there is no universal “trust color” or “buy color”. A red button works when red means “action” on your page and appears nowhere else. The same red sprinkled across warnings, headings and decorations means nothing in particular.
Do this: pick one color to mean “click here” and use it only for that.
Color and brand personality
If color does not have fixed effects on behavior, does it at least shape how a brand comes across? Here the evidence is firmer, if more modest in what it claims.
Lauren Labrecque and George Milne's “Exciting red and competent blue” (Journal of the Academy of Marketing Science, 2012) ran four studies. They mapped hues onto brand personality dimensions, tested whether saturation (how intense a color is) and value (how light or dark) amplify those traits, showed that changing color could shift perceived brand personality and purchase intent, and looked at how color affects a brand's likability and familiarity. The title gives away the headline associations: red with excitement, blue with competence.
Two things to keep in mind. First, these are people's ratings of logos and brands in controlled studies, not sales data from live websites. Second, the finding is about fit: color helps when it matches the personality you want to project. A children's music school and a tax adviser should not share a palette, but neither has a “correct” one.
Do this: write down three adjectives you want visitors to use about your business. Then ask whether your colors say the same thing. Saturated, warm colors tend to read as lively; darker, cooler, quieter colors tend to read as serious.
Same color, different countries
Many color guides include a table of what each color “means” in China, India or Brazil. Treat those tables with care. The research shows overlap as well as differences, and the differences are rarely as tidy as the tables suggest.
Thomas Madden, Kelly Hewett and Martin Roth's 2000 study in the Journal of International Marketing asked people in eight countries about color preferences and meanings, including which colors they would combine in a logo. Its conclusion: cross-cultural patterns of both similarity and dissimilarity.
The largest study to date is newer. Domicele Jonauskaite and colleagues, in Psychological Science (2020), asked 4,598 people from 30 nations, speaking 22 native languages, to link 20 emotion concepts to 12 color terms. The associations were strikingly similar worldwide, with an average similarity of r = .88. Yet nationality still predicted associations beyond the universal pattern, and countries that were linguistically or geographically closer were more alike.
Note what was tested: color words, not color samples on a screen. Still, the practical lesson is sound. If you sell across borders, broad associations travel reasonably well, but test specific choices, such as a color linked to mourning or a political party, with people from the market you are entering.
The 62–90% myth
You have probably seen it: “People make up their minds within 90 seconds, and 62–90% of that assessment is based on color alone.” It appears in countless infographics, always credited to the same source.
That source is Satyendra Singh's “Impact of color on marketing” (Management Decision, 2006), a seven-page literature review. We read the full paper. The sentence appears in the abstract. The body of the paper does not repeat it, and no study, data or citation is given for it anywhere in the text.
That does not make color unimportant. It means nobody has shown that color accounts for most of a first impression, and the number should not drive design decisions. (First impressions are covered properly, with their own myths, in First impressions.)
Do this: when a statistic about design sounds too clean, look for the method. If you cannot find who was tested and what was measured, do not build on it.
One accent, used sparingly
If you remember one thing from this chapter, make it this: attention goes to what is different. That idea has a name and a history.
In her 1933 dissertation research in Berlin, von Restorff showed that an item that stood out from an otherwise similar list (for example, one syllable among numbers) was recalled better. Her lesser-known point, highlighted by R. Reed Hunt in 1995, is that difference only works against a background of similarity. Add more “different” items and the advantage fades.
Hunt's “What von Restorff really did” is worth knowing because it corrects the usual summary. Von Restorff argued that an isolated item is not remembered better simply because it is different; it works because everything else is alike. She reasoned through the extreme case: keep swapping the other items for unrelated things until every item differs from every other. The original isolate is then just as different as before, yet it was remembered better in the list where everything else was alike. Sameness, not difference alone, did the work.
This is why pages with a single accent color feel calm and clear, and pages like Ana's feel noisy. Every extra accent color dilutes the first. For the attention side of this, see Attention and whitespace and the banner blindness tip.
And the 60-30-10 rule?
Designers often quote a “60-30-10 rule”: 60% dominant color, 30% secondary, 10% accent. It comes from interior decorating, and we found no research behind the specific numbers. It is a reasonable heuristic because it builds in exactly what von Restorff described: lots of sameness, a little contrast, a sliver of accent. Use it as a starting point, not a law. On many good landing pages the accent is closer to 2% than 10%.
Contrast is not a matter of taste
Ana's pale orange timetable failed for a reason that has nothing to do with emotion: there was not enough difference in lightness between the text and the background. This is the one part of color on the web with clear, testable rules.
1.4.3 Contrast (Minimum): text and images of text need a contrast ratio of at least 4.5:1, or 3:1 for large text (at least 18 point, or 14 point bold, which is roughly 24 px or 18.7 px). Logos and purely decorative text are exempt. 1.4.11 Non-text Contrast: the visual information needed to identify interface components and their states, such as a form field's border or a focus outline, and parts of graphics needed to understand them, need at least 3:1 against adjacent colors.
Source: W3C, WCAG 2.2. AA is the level most organizations aim for. General information, not legal advice. Which accessibility laws apply to you is covered in Accessibility as conversion.
L1 and L2 are the relative luminance of the lighter and darker color, a weighted mix of red, green and blue in which green counts most and blue least. The ratio runs from 1:1 (identical) to 21:1 (black on white).
Some results surprise people. Pure red (#ff0000) on white comes out at 4.0:1, which fails for body text. The lightest neutral grey that passes on white is about #767676 (4.54:1); one step lighter, #777777, falls just short. Fashionable light-grey text such as #999999 manages only 2.85:1. Try your own colors.
Book your first class
Mondays and Thursdays at 18:30. Mats provided, no experience needed.
- Body text, AA (4.5:1)Fail
- Large text, AA (3:1)Fail
- Icons, borders, focus rings, AA (3:1)Fail
- Body text, AAA (7:1)Fail
What comes after 4.5:1
The WCAG 2 formula has known weaknesses. It treats a thin 14-pixel line and a heavy 40-pixel headline almost the same, and it can misjudge some combinations, especially light text on dark or saturated backgrounds. An alternative, the Advanced Perceptual Contrast Algorithm (APCA), weighs font size and weight and has many supporters.
Do not switch yet. APCA appeared in early drafts of WCAG 3 but, as Adrian Roselli documents, it was removed from the working draft in July 2023. The current WCAG 3.0 Working Draft of 10 September 2026 still says the contrast algorithm is to be determined, assumes it will include a size and weight factor, and notes that a separate requirement may be needed for red/green color vision deficiency. WCAG 3 is a work in progress; WCAG 2.2 remains the reference. Use APCA as a second opinion if you like, but meet 4.5:1 first.
Not everyone sees your red
About 1 in 12 men have a color vision deficiency, according to the US National Eye Institute, most commonly difficulty telling red from green. For women the figure is around 1 in 200, according to Colour Blind Awareness. Rates vary between populations, but if a few hundred men visit your site each month, dozens of them see your colors differently from you.
The classic failures are red error borders, green “success” states, and charts where the only difference between two lines is red against green. Switch the view below to see how a typical form and chart look under three common types of color blindness.
Color only
EmailPhoneColor plus text
EmailPhone✕ Please add a phone number with country codeSimulation matrices from Machado, Oliveira & Fernandes (2009), severity 1.0 (dichromacy). An approximation: real perception varies from person to person.
Color may not be the only visual means of conveying information, indicating an action, prompting a response or distinguishing a visual element. A red border needs an error message; a link inside a paragraph needs an underline or another non-color cue; a chart needs labels, patterns or direct annotation.
Source: W3C, WCAG 2.2, SC 1.4.1. Level A is the baseline tier. Not legal advice.
Do this: take a screenshot of your page and turn it greyscale. Anything you can no longer understand, such as which field has an error or which plan is highlighted, needs a second cue: text, an icon, an underline or a pattern.
Can you own a color?
Sometimes, yes, and that matters if you are tempted to borrow a famous brand's signature shade. Courts have allowed single colors to be registered as trade marks, but they set the bar high.
The Dutch telecoms company Libertel tried to register the color orange for telecom goods and services. The EU Court of Justice held that a color per se can have distinctive character, but it must be represented clearly and precisely: a paper sample alone is not enough, while an internationally recognized color code may be. Distinctiveness without prior use will be rare, though a color can acquire it through use. Registries must also weigh the general interest in not unduly restricting the availability of colors for other traders.
Source: Judgment in C-104/01, EUR-Lex. Not legal advice.
Cadbury's purple: not precise enough
Cadbury applied in the UK to register a purple (Pantone 2685C) applied to the whole visible surface of chocolate packaging, or being its “predominant” color. Nestlé opposed. In Société des Produits Nestlé v Cadbury UK [2013] EWCA Civ 1174, decided on 4 October 2013, the Court of Appeal held that the word “predominant” opened the door to many different visual forms, so the description was not a registrable sign, and the lower decisions allowing registration were wrong.
England and Wales Court of Appeal; following Libertel and Heidelberger Bauchemie (C-49/02).
Magenta against plum
In T-Mobile US v. Aio Wireless, 991 F. Supp. 2d 888 (S.D. Tex. 2014), T-Mobile and its parent Deutsche Telekom won a preliminary injunction on 3 February 2014 stopping AT&T's new Aio Wireless brand from using a plum color as a central part of its trade dress, against T-Mobile's magenta.
US District Court, Southern District of Texas, Judge Lee H. Rosenthal. A preliminary ruling in a crowded market for the same services.
For a small business, the practical risk is narrow. You can use purple, orange or magenta freely. Trouble starts when you adopt a competitor's signature color in the same market, in a way that could make customers think you are connected. Choose a palette for your brand's personality, not to look like the market leader.
General information, not legal advice. Trade mark protection depends on registration, use and jurisdiction; check with a lawyer before building a brand around a color associated with someone else.
What to change tomorrow
Ana rebuilt her palette in an hour: warm off-white, near-black text, one muted green for every “Book” button and nothing else. The timetable became readable in sunlight. The site stopped looking like a sale.
Give one color one job
Choose a single accent for your primary action. Remove it from headings, icons and decorations so it keeps its meaning.
Check every text color
Run body text, links, buttons, placeholders and captions through a contrast checker. Body text needs 4.5:1; large text and interface borders need 3:1.
Add a second cue
Errors get words and an icon, links get underlines, charts get labels. Then check your page in greyscale and with a color-blindness simulator.
Match palette to personality
Pick colors that fit the three adjectives you want customers to use, and test them on a few people from your market rather than trusting a color-meaning table.
Drop the myths
Delete any decision that rests on “red converts”, “blue builds trust” or “90% of judgments are color”. Test what matters on your own page.
What the research supports
Color associations are broadly shared across cultures but not identical, and color can shift how a brand's personality is perceived in controlled studies. The famous claim that red impairs test performance has not survived meta-analysis, and the “62–90% of judgments” figure has no traceable source. The firmest ground is not psychological at all: contrast ratios and non-color cues are measurable, standardized, and directly affect whether people can read and use your page.
Sources
- Elliot, A. J. & Maier, M. A. (2012). Color-in-Context Theory. Advances in Experimental Social Psychology, 45, 61–125. Bibliographic record verified; the theory's core claims as summarized in the literature cited below.
- Elliot, A. J., Maier, M. A., Moller, A. C., Friedman, R. & Meinhardt, J. (2007). Color and psychological functioning: The effect of red on performance attainment. Journal of Experimental Psychology: General, 136(1), 154–168. Abstract-level findings.
- Gnambs, T. (2020). Limited evidence for the effect of red color on cognitive performance: A meta-analysis. Psychonomic Bulletin & Review, 27(6), 1374–1382. Full text consulted.
- Labrecque, L. I. & Milne, G. R. (2012). Exciting red and competent blue: the importance of color in marketing. Journal of the Academy of Marketing Science, 40(5), 711–727 (online 2011). Abstract consulted; full text not accessible to us, so we describe only the study aims and headline associations.
- Madden, T. J., Hewett, K. & Roth, M. S. (2000). Managing Images in Different Cultures: A Cross-National Study of Color Meanings and Preferences. Journal of International Marketing, 8(4), 90–107. Abstract consulted.
- Jonauskaite, D. et al. (2020). Universal Patterns in Color-Emotion Associations Are Further Shaped by Linguistic and Geographic Proximity. Psychological Science, 31(10), 1245–1260. Abstract consulted for sample, method and results.
- Singh, S. (2006). Impact of color on marketing. Management Decision, 44(6), 783–789. Full text read to trace the 62–90% claim.
- von Restorff, H. (1933). Über die Wirkung von Bereichsbildungen im Spurenfeld. Psychologische Forschung, 18, 299–342. Described via Hunt (1995).
- Hunt, R. R. (1995). The subtlety of distinctiveness: What von Restorff really did. Psychonomic Bulletin & Review, 2(1), 105–112. Full text consulted.
- W3C (2023). WCAG 2.2, SC 1.4.1, 1.4.3, 1.4.11 and the definitions of large-scale text and relative luminance. W3C (2026). WCAG 3.0 Working Draft, 10 September 2026, text contrast requirements and editor's notes.
- Roselli, A. (2026). WCAG3 Contrast as of April 2026. History of APCA's removal from the July 2023 draft.
- National Eye Institute. Color Blindness; Colour Blind Awareness. About colour blindness. Prevalence figures.
- Machado, G. M., Oliveira, M. M. & Fernandes, L. A. F. (2009). A Physiologically-based Model for Simulation of Color Vision Deficiency. IEEE TVCG, 15(6), 1291–1298. Simulation matrices used in the demo.
- CJEU, Case C-104/01, Libertel Groep BV v Benelux-Merkenbureau, 6 May 2003. EUR-Lex.
- Société des Produits Nestlé SA v Cadbury UK Ltd [2013] EWCA Civ 1174. The National Archives, Find Case Law.
- T-Mobile US, Inc. v. Aio Wireless LLC, 991 F. Supp. 2d 888 (S.D. Tex. 2014). CourtListener record; outcome confirmed via Jones Day, counsel for Deutsche Telekom.
Sources checked 18 October 2026. Contrast ratios in the text were calculated with the WCAG 2.2 relative luminance formula. Ana and her studio are fictional; all figures and demos are original illustrations.