The chart that assigns a meaning to each color is not a research finding. It is a genre. And the researchers who built the field are unusually direct about this: the leading review of color psychology, published in the Annual Review of Psychology, concludes that “the literature remains at a nascent stage of development” and that “considerable work on boundary conditions, moderators, and real-world generalizability is needed before strong conceptual statements and recommendations for application are warranted.”

That sentence is from the people best placed to make those recommendations. They declined to.

Which leaves a practical question: if color meaning is not established, what is? The answer is more useful than the chart. Three things are genuinely settled, one is conditionally supported, and the rest is open. This article sorts them.

What the field says about itself

Start with the assessment written by the field’s own lead researcher, because it is more critical than anything a marketing article would say.

On the state of the evidence. The 2014 review concludes that color “can carry important meaning and can have an important impact on people’s affect, cognition, and behavior”, and immediately qualifies it: the literature “remains at a nascent stage of development.”

On measurement. A 2015 open-access assessment notes that “in most research it (still) is not” the case that color properties are controlled at the spectral level. Control “is typically done improperly at the device (rather than the spectral) level, is impossible to implement (e.g., in web-based platform studies), or is ignored altogether.” The consequence: such work “is inherently fraught with interpretational ambiguity.”

On statistical power. “Much of the research remains underpowered”, and “underpowered samples commonly produce overestimated effect size estimates”. The recommendation is that “findings from small sampled studies should be considered preliminary.”

On the temptation. The same paper names the mechanism that produces color meaning charts: “findings from color research can be provocative and media friendly, and the public (and the field as well) can be tempted to reach conclusions before the science is fully in place.”

And the conclusion it reaches. “It is premature to offer any bold theoretical statements, definitive empirical pronouncements, or impassioned calls for application.”

One more marker, from a review of color work in marketing specifically. Its opening observes that many disciplines have gained insight into color perception, “yet there exists relatively little research in the field of marketing.” The gap is not hidden. It is stated at the top of the literature.

Structural weaknesses of the color psychology literature as stated by its own researchersStructural weaknesses of the color psychology research literature, as identified not by critics of the field but by its own leading researchers in peer reviewed assessments. The first weakness concerns measurement, since although some research in the area controls color properties at the spectral level, in most research it still is not controlled, with control typically done improperly at the device rather than the spectral level, impossible to implement in web based platform studies, or ignored altogether, and the stated consequence is that findings from uncontrolled research can be informative in initial explorations of color hypotheses but such work is inherently fraught with interpretational ambiguity. The second weakness concerns statistical power, since although some research uses large and fully powered samples, much of the research remains underpowered, which is a particular problem when the initial demonstration of an effect is underpowered, because underpowered samples commonly produce overestimated effect size estimates and basing subsequent sample sizes on those estimates perpetuates the problem, so that findings from small sampled studies should be considered preliminary. The third weakness concerns the gap between findings and application, with the leading review of the field concluding that the literature remains at a nascent stage of development and that considerable work on boundary conditions, moderators and real world generalizability is needed before strong conceptual statements and recommendations for application are warranted, and a subsequent assessment stating that it is premature to offer any bold theoretical statements, definitive empirical pronouncements or impassioned calls for application. The same assessment identifies the mechanism that produces popular color meaning charts, noting that findings from color research can be provocative and media friendly and that the public, and the field as well, can be tempted to reach conclusions before the science is fully in place.What the researchers say about their own fieldMeasurementColor is mostly not controlledat the spectral level.”inherently fraught withinterpretational ambiguity”Statistical power”Much of the researchremains underpowered.”Underpowered samplesoverestimate effect sizes.Application”A nascent stage ofdevelopment.""Premature to offer …calls for application.”And the sentence that explains the entire color meaning chart industry”Findings from color research can be provocative and media friendly, and the public (and the field as well) can betempted to reach conclusions before the science is fully in place.”This is not a critique from outside the field. It is the position of the researchers who built it, in peer reviewed venues,declining to make the recommendations that marketing articles make on their behalf.
Three structural weaknesses, named by the field's own researchers rather than by critics of it. Source : Elliot, Color and psychological functioning, Frontiers in Psychology 6:368, 2015 (2015)

The most studied color effect, after replication

If any color effect were solid, it would be the most researched one. It is worth looking at what happened when it was tested properly.

A 2018 meta-analysis of the red and attractiveness effect. For men rating women, “a small, statistically significant effect (d = 0.26 [0.12, 0.40], p = .0004, N = 2,961), with substantial heterogeneity”. For women rating men, “a very small effect (d = 0.13 [0.01, 0.25], p = .03, N = 2,739)”, with “evidence of upward bias in the estimate.”

And two findings inside it that matter more than the effect size. Effect sizes “may have declined over time”, and “may be smaller in preregistered studies”. The second is the standard signature of publication bias.

A preregistered replication, run the careful way. The authors “obtained original materials, planned for informative sample sizes, pre-registered our study, used a positive control, and adopted quality controls.” The result: “For men, we found a very weak effect in the predicted direction (d = 0.09, 95% CI [-0.17, 0.34], N = 242). For women, we found a very weak effect in the opposite direction (d = -0.09, 95% CI [-0.30, 0.12], N = 360).”

Both confidence intervals contain zero. The honest reading is not that the effect is absent, but that it is small, unstable, and smaller the more carefully it is measured.

One boundary to state plainly. This work concerns red and romantic attractiveness, not brand color. It is not a replication of any branding study. It is evidence about how robust the wider field is, and it should be cited that way and no further.

Which surfaces the more telling gap. No meta-analysis or systematic replication effort was found addressing color effects in marketing or branding specifically. The area that generates the most confident advice has produced the least verification of it.

Effect sizes for the red and attractiveness effect in meta analysis and in preregistered replicationEffect sizes reported for the most extensively studied effect in color psychology, namely the effect of the color red on perceived romantic attractiveness, comparing the pooled meta analytic estimate against a preregistered replication conducted with tightened methods. The 2018 meta analysis reported, for men rating women, a small statistically significant effect of zero point two six standard deviations with a confidence interval from zero point one two to zero point four zero, a probability value of zero point zero zero zero four and a combined sample of two thousand nine hundred and sixty one, accompanied by substantial heterogeneity and equivocal results regarding the possibility of upward bias in the estimate. For women rating men it reported a very small effect of zero point one three with a confidence interval from zero point zero one to zero point two five, a probability value of zero point zero three and a combined sample of two thousand seven hundred and thirty nine, with evidence of upward bias in the estimate. Moderator analyses in the same meta analysis suggested that effect sizes may have declined over time for both genders and may be smaller in preregistered studies, the latter being the standard signature of publication bias. The 2017 preregistered replication, in which the authors obtained the original materials, planned for informative sample sizes, preregistered the study, used a positive control and adopted quality controls, found for men a very weak effect in the predicted direction of zero point zero nine with a ninety five percent confidence interval from negative zero point one seven to zero point three four in a sample of two hundred and forty two, and for women a very weak effect in the opposite direction of negative zero point zero nine with a confidence interval from negative zero point three zero to zero point one two in a sample of three hundred and sixty. Both replication confidence intervals contain zero. An essential boundary on this evidence is that it concerns red and romantic attractiveness rather than brand color, and it is not a replication of any branding study, so it speaks to the robustness of the wider color psychology field and should not be cited further than that.The best-studied color effect, measured twiceMeta-analysis, 2018Men rating womend = 0.26N = 2,961Women rating mend = 0.13N = 2,739Preregistered replication, 2017Men rating womend = 0.09N = 242Women rating mend = -0.09N = 360Both 95% confidence intervals contain zero. The second is in the opposite direction.Two findings inside the meta-analysisEffects “may have declined over time” and”may be smaller in preregistered studies”.The boundary on citing thisThis is red and romantic attractiveness, notbrand color. It is evidence about the field.
The best-studied color effect shrinks as the method tightens. Both preregistered confidence intervals contain zero. Source : Lehmann, Elliot and Calin-Jageman, Evolutionary Psychology 2018; Lehmann and Calin-Jageman, Social Psychology 2017 (2018)

What does hold up: congruence, not meaning

There is a defensible finding in the marketing literature, and it is a different shape from the one the charts describe.

The study. Two experiments on color and brand logo appropriateness, published in 2006.

What experiment one found. It “explores the notion of congruity, showing that it is more appropriate for functional products to be presented in functional colors, and sensory-social products in sensory-social colors.”

What experiment two added. “When people know how brands are attempting to position themselves, people consider colors congruent with those positions to be more appropriate.”

Read that clause carefully. The judgement is conditional on the audience already knowing the positioning. Color is being evaluated against a claim, not delivering one. That is the opposite of the chart’s premise.

Which is a workable principle. Choose color that fits the position you are already communicating, in your category, for your audience. Not color chosen to install a feeling in someone who knows nothing about you.

And a boundary worth keeping. Two experiments, published in 2006. The full text is behind a paywall and the authors’ own stated limitations were not accessible for this article. Treat it as a reasonable working principle, not as a settled law.

Congruence as a supported finding compared with intrinsic color meaning as an unsupported oneContrast between the claim that colors carry intrinsic meanings, which the research does not establish, and the claim of congruence, which a published marketing study does support in a conditional form. The intrinsic meaning claim holds that a given color installs a given feeling or attribute in any viewer regardless of context, so that choosing the color is itself the act of communication, and this is the premise of the color meaning charts widely circulated in marketing material. The congruence claim, examined in a two experiment study published in 2006, runs in the opposite direction. The first experiment explored the notion of congruity and found that it is more appropriate for functional products to be presented in functional colors and for sensory social products to be presented in sensory social colors, making appropriateness a function of product category rather than of colour alone. The second experiment examined the effect of red and blue on brands of products classifiable as both functional and sensory social, and found that when people know how brands are attempting to position themselves, people consider colors congruent with those positions to be more appropriate. The conditional clause is the load bearing part, because the judgement depends on the audience already knowing the positioning, which means colour is being evaluated against a claim rather than delivering one. The workable principle that follows is to choose colour that fits the position already being communicated, within the relevant category and for the relevant audience, rather than choosing colour intended to install a feeling in somebody who knows nothing about the brand. The stated boundary is that this rests on two experiments published in 2006 whose full text is paywalled and whose authors’ own declared limitations were not accessible, so it should be treated as a reasonable working principle rather than as settled law.Two different claims, often confusedIntrinsic meaningThe color installs a feeling inany viewer, regardless of context.Choosing the color is the act ofcommunication. Not established.CongruenceThe color is judged appropriateagainst a known position.The color is evaluated against aclaim, not delivering one.The clause that carries the whole finding”When people know how brands are attempting to position themselves, people consider colors congruentwith those positions to be more appropriate.” Two experiments, 2006. A working principle, not a law.So: fit the position you already communicate. Do not expect color to communicate it to a stranger.
The supported finding runs the opposite way from the chart: color is judged against a position the audience already knows. Source : Bottomley and Doyle, Marketing Theory 6(1), 2006 (2006)

The only numbers in the whole subject

Everything above is contested. This part is a published specification with exact thresholds, and it constrains a palette more than any meaning ever will.

Normal text, level AA. “The visual presentation of text and images of text has a contrast ratio of at least 4.5:1.”

Large text, level AA. 3:1. And large is defined precisely: “at least 18 point or 14 point bold”, with the standard noting that for many mainstream body fonts this is “roughly equivalent to 1.2 and 1.5 em”, while warning that “authors would need to check this for the particular fonts in use.”

Level AAA, if you are aiming higher. 7:1 for normal text and 4.5:1 for large text.

Interface components and meaningful graphics. 3:1 against adjacent colors, covering “visual information required to identify user interface components and states” and “parts of graphics required to understand the content.” This is the criterion that governs your button borders, your form field outlines and your chart colors.

And the exemption almost nobody knows. “Text that is part of a logo or brand name has no contrast requirement.” Your logo is exempt. Everything the logo’s palette then gets used for is not.

One criterion sits below all of these, at level A. “Color is not used as the only visual means of conveying information, indicating an action, prompting a response, or distinguishing a visual element.” A status shown only in red and green fails the lowest bar in the standard.

Contrast ratio requirements applying to a brand color paletteThe contrast ratio thresholds that a brand color palette must clear under the web content accessibility guidelines version two point two, published as a recommendation in December 2024, which are the only numeric requirements in the entire subject of brand color. At level double A, the visual presentation of text and images of text must have a contrast ratio of at least four point five to one, except that large scale text and images of large scale text require at least three to one. Large scale is defined in the glossary as at least eighteen point, or fourteen point bold, or a font size yielding equivalent size for Chinese, Japanese and Korean fonts, with a note stating that for many mainstream body text fonts fourteen and eighteen point is roughly equivalent to one point two and one point five em, or one hundred and twenty and one hundred and fifty percent of the default body size, while cautioning that authors would need to check this for the particular fonts in use. At level triple A the same thresholds rise to seven to one for normal text and four point five to one for large text. A separate level double A criterion requires a contrast ratio of at least three to one against adjacent colors for the visual information required to identify user interface components and their states, excepting inactive components or those whose appearance is determined by the user agent, and for parts of graphics required to understand the content, excepting where a particular presentation of graphics is essential to the information conveyed. Two exemptions apply to the text criteria, namely incidental text that is part of an inactive component, pure decoration, invisible, or part of a picture containing significant other visual content, and logotypes, where text that is part of a logo or brand name has no contrast requirement at all. Beneath all of these sits a level A criterion, the lowest conformance level in the standard, stating that color is not to be used as the only visual means of conveying information, indicating an action, prompting a response or distinguishing a visual element, which means a status communicated solely through red and green fails the standard’s minimum bar.The thresholds, exactlyWhat is being coloredLevel AALevel AAANormal text4.5:17:1Large text18pt, or 14pt bold3:14.5:1UI components and meaningful graphicsButton borders, field outlines, chart colors3:1sameText that is part of a logo or brand nameNo requirementAnd one rule beneath all of them, at level A”Color is not used as the only visual means of conveying information, indicating an action, prompting a response,or distinguishing a visual element.” A status shown only in red and green fails the standard’s lowest level.The logo is exempt. Everything the logo’s palette is then reused for is not.
Exact ratios, with the logo exemption and the level A rule that governs status colors. Source : W3C, Web Content Accessibility Guidelines 2.2, 12 December 2024 (2024)

The constraint that shows up after approval

A palette is approved on one screen, in one theme, and then has to survive three rendering conditions it was never tested against.

Dark mode is a second palette, not an inversion. The contrast thresholds apply in both themes independently. A pairing that clears 4.5:1 on white will not automatically clear it on a dark ground, and the failure appears in the theme nobody reviewed.

Print does not use your screen colors. The web palette is defined in RGB, and the image format most brand assets ship in cannot express anything else: the specification for it forecloses redefining its channels to mean process color. Print output relies on a declared destination profile in the PDF, which “describes the final destination device you’ll use to reproduce the color”, and that conversion is where an approved screen color becomes a different printed one.

And the disclosure case is regulated, not aesthetic. Where a claim requires a disclosure, US advertising guidance is explicit that “the size, color, and graphics of the disclosure affect its prominence”, that “a disclosure in a color that contrasts with the background emphasizes the text”, and that “information in a color that blends in with the background of the ad is likely to be missed.” A palette that renders fine print in a low-contrast tint is a compliance question, not a design preference.

Which suggests one test before approval. Take the two or three pairings the brand will actually use most, and check them in light theme, dark theme and print export. Most palettes fail at least one, and the fix is cheap only before the guidelines are published.

A color can be owned, on one condition

If color meaning is unsettled, color ownership is not. The Supreme Court decided it in 1995, unanimously.

The holding. “The Lanham Act permits the registration of a trademark that consists, purely and simply, of a color.”

And the reasoning, which is the useful part for a brand team. “The imaginary word ‘Suntost,’ or the words ‘Suntost Marmalade,’ on a jar of orange jam immediately would signal a brand or a product ‘source’; the jam’s orange color does not do so. But, over time, customers may come to treat a particular color on a product or its packaging … as signifying a brand.”

Which sets the condition. Registration is available “where that color has attained ‘secondary meaning’ and therefore identifies and distinguishes a particular brand.”

And the trademark office states the consequence bluntly. Color marks are never inherently distinctive, so an examining attorney must refuse registration on the principal register unless acquired distinctiveness is established, and “the burden of proving that a color mark has acquired distinctiveness is substantial.” A statement of long use is not enough; the applicant “must demonstrate that the color has acquired source-indicating significance in the minds of consumers.”

One distinction that has moved recently. The office’s own current guidance separates a single color mark, which is never inherently distinctive, from a multiple color mark on product packaging, which may be inherently distinctive depending on the character of the color design. That change followed a 2020 Federal Circuit decision.

What all of this says to a brand. A color becomes yours through consistent, exclusive, prolonged use in one category. Not by choosing it well. Ownership is an outcome of repetition, which is also, incidentally, the only mechanism by which a color comes to mean anything at all.

How a color qualifies for trademark registration in the United StatesHow a color qualifies for trademark registration in the United States, following the unanimous Supreme Court decision of 1995 and the trademark office’s subsequent examination guidance. The court held that the trademark statute permits the registration of a trademark that consists, purely and simply, of a color. Its reasoning distinguished an invented word placed on a jar of orange jam, which would immediately signal a brand or product source, from the jam’s orange color, which would not do so, while observing that over time customers may come to treat a particular color on a product or its packaging as signifying a brand, giving the examples of pink on a firm’s insulating material and red on the head of a large industrial bolt. The court concluded that there is no obvious theoretical objection to the use of color alone as a trademark where that color has attained secondary meaning and therefore identifies and distinguishes a particular brand, with secondary meaning defined as acquired when in the minds of the public the primary significance of a product feature is to identify the source of the product rather than the product itself. The examination consequence is that color marks are never inherently distinctive, so the examining attorney must refuse registration on the principal register unless the applicant establishes acquired distinctiveness, and the office states that the burden of proving a color mark has acquired distinctiveness is substantial, that a mere statement of long use is not sufficient, and that the applicant must demonstrate the color has acquired source indicating significance in the minds of consumers. Current office guidance separates a single color mark, which is never inherently distinctive, from a multiple color mark used on product packaging, which may be inherently distinctive depending on the character of the color design, a distinction that followed a 2020 Federal Circuit decision. The practical implication for a brand is that a color becomes ownable through consistent, exclusive and prolonged use within one category rather than through being well chosen, so ownership is an outcome of repetition.A color can be owned. Here is the condition.”The Lanham Act permits the registration of a trademark that consists, purely and simply, of a color.” (1995, unanimous)But never on day oneColor marks are never inherentlydistinctive. The examiner must refuseunless acquired distinctiveness isestablished.What has to be shownThat the color “has acquired source-indicating significance in the minds ofconsumers”. A statement of long useis not sufficient.One distinction in current guidance: a single color mark is never inherently distinctive, while a multiple color mark onproduct packaging may be, depending on the character of the color design.What that tells a brand teamA color becomes yours through consistent, exclusive, prolonged use in one category. Ownership is an outcome of repetition.
Color is registrable, but never on its own merits. It has to have become a source indicator through use. Source : Qualitex Co. v. Jacobson Products Co., 514 U.S. 159 (1995), and USPTO examination guidance (1995)

What to do with this

Stop asking what a color means and start asking three answerable questions. Is it congruent with the position you already communicate, in your category? Does every text and interface pairing clear 4.5:1, or 3:1 where the standard allows it? And is any information in your product carried by color alone?

Then choose on grounds that are actually yours: distinctiveness against the competitors in your category, reproduction across screen and print, and whether you can commit to it for long enough that repetition does the work no palette theory can. Congruence is the one criterion a palette cannot answer on its own, because it depends on a position fixed beforehand, and settling the positioning before the visual identity is drawn is what stops the two contradicting each other.

If somebody presents a color meaning chart during the process, the useful response is not to argue about blue. It is to note that the researchers who study this decline to make those recommendations, and to move the conversation to the three questions above.

For the palette’s typographic partner, see brand typeface licensing explained. For the files that carry the color into production, logo file formats and when to use them.