Short answer
It depends on what the data is. For shares of named colours, the usual figure is the achromatic share, but it hides how the market is spread, so it should be read with a concentration measure and an entropy measure. For data with real colour values, chroma statistics are possible and the named-colour measures are not. The measures regularly disagree: in UK registrations the three leading colours gained about three points in four years while the neutral share barely moved. A claim that something is 'less colourful' means little until it says which measure moved.
The achromatic share is the percentage filed under white, black, grey or silver; the chromatic share is the rest. It is the figure behind 'four in five cars are neutral', and it has two blind spots. It cannot see inside the neutral total: 80% spread over four colours and 80% in one colour score the same. And it takes the source's categories on trust, so a warm grey-brown filed under grey counts as achromatic and the same paint filed under brown does not. In Germany's registrations it rises from 74.6% to 78.8% between 2014 and 2023. Because the scale flattens near the top, the chromatic share is the better way to feel a change: that is a fall from 25.4% to 21.2%, about a sixth of all coloured cars.
Top-three concentration is the combined share of the three largest named colours, whatever they are. It answers a different question from the achromatic share and the two can part company. In UK registrations the three leaders held 62.5% in 2022 and 65.8% in 2025, a clear rise, while the achromatic share moved only from 68.6% to 70.0%. The reason is in the names: the top three were grey, black, white in 2022 and grey, black, blue in 2025. A chromatic colour entered the podium as white fell, so the market grew more concentrated without growing more neutral. The measure ignores everything below third place and rises whenever a source merges two categories.
Shannon entropy, H = −Σ p ln p over the shares p, measures how evenly choices are spread across all categories at once. It is zero when one colour takes everything and ln k when k colours are equal. Two transformations make it readable. Dividing by ln k gives a normalised entropy between 0 and 1, relative to the most even split the source's own categories allow. Taking exp(H) gives the effective number of colours: the number of equally popular colours that would be this evenly spread. Six categories at 16.7% each have an effective number of six; a market of 60%, 30% and ten colours at 1% has about three. This site's older concentration table uses 1 ÷ Σp², which leans harder on the leaders and always gives a smaller number; both are reported, and they should not be mixed in one column.
Entropy counts categories, not colours. A source with fourteen can score higher than one with seven describing the same cars. Normalising helps and brings its own trap: the divisor changes whenever a category empties. The Dutch register shows it. Its effective number of colours barely moves between 2010 and 2025 (5.7 and 5.6), yet its normalised entropy shifts from 0.66 to 0.69 because the number of categories in use went from 14 to 12 as cream and pink emptied. A remainder bucket is a further distortion, since 'other' is counted as one colour however many it holds. Entropy is trustworthy within one source whose categories held still, and a rough guide anywhere else.
None of the above measures colour. They measure filing. Where each item has a colour value, chroma can be calculated directly: the mean and median chroma of the items, and the share below a stated threshold. In the LEGO catalogue, the mean chroma of an opaque part falls from 50.9 in 1975 to 25.8 in 2025 in CIELAB units, and the share of parts below a chroma of 30, which Colourwise calls muted, rises from 35.0% to 62.8%. These figures cannot be produced for cars, because a registry records a word and not a value: there is no chroma of 'grey'. Equally, an achromatic share cannot be read off a measured palette without choosing a threshold, and the choice is the result. The two families of measure are kept in separate modules that do not accept each other's data.
| Source | Year | Achromatic (%) | Chromatic (%) | Top three (%) | Entropy (bits) | Normalised entropy | Effective colours, exp(H) | Categories |
|---|---|---|---|---|---|---|---|---|
| Germany (KBA registrations) | 2014 | 74.6 | 25.4 | 74.6 (black, grey, white) | 2.43 | 0.87 | 5.4 | 7 |
| Germany (KBA registrations) | 2023 | 78.8 | 21.2 | 78.8 (grey, black, white) | 2.37 | 0.84 | 5.2 | 7 |
| United Kingdom (SMMT registrations) | 2022 | 68.6 | 31.4 | 62.5 (grey, black, white) | 2.78 | 0.80 | 6.9 | 11 |
| United Kingdom (SMMT registrations) | 2025 | 70.0 | 30.0 | 65.8 (grey, black, blue) | 2.75 | 0.79 | 6.7 | 11 |
| Netherlands (RDW register, survivors) | 2010 | 73.8 | 26.2 | 74.7 (grey, black, blue) | 2.51 | 0.66 | 5.7 | 14 |
| Netherlands (RDW register, survivors) | 2025 | 73.1 | 26.9 | 73.1 (grey, black, white) | 2.49 | 0.69 | 5.6 | 12 |
| Global (DuPont 2009 report) | 2009 | 77.0 | 23.0 | 64.0 (silver, black, white) | 2.72 | 0.86 | 6.6 | 9 |
| Global (DuPont 2010 report) | 2010 | 82.8 | 17.2 | 66.7 (silver, black, white) | 2.65 | 0.84 | 6.3 | 9 |
| Global (DuPont 2011 report) | 2011 | 77.0 | 23.0 | 64.0 (white, silver, black) | 2.83 | 0.85 | 7.1 | 10 |
| Global (DuPont 2012 report) | 2012 | 76.0 | 24.0 | 62.0 (white, black, silver) | 2.83 | 0.85 | 7.1 | 10 |
| Global (BASF 2019 report) | 2019 | 78.0 | 22.0 | 68.0 (white, black, grey) | 2.60 | 0.78 | 6.1 | 10 |
| Global (BASF 2021 report) | 2021 | 77.0 | 23.0 | 69.0 (white, black, grey) | 2.67 | 0.80 | 6.4 | 10 |
| Global (BASF 2022 report) | 2022 | 81.0 | 19.0 | 73.0 (white, black, grey) | 2.58 | 0.74 | 6.0 | 11 |
| Global (Axalta 2023 report) | 2023 | 81.0 | 19.0 | 71.0 (white, grey, black) | 2.68 | 0.81 | 6.4 | 10 |
| Global (Axalta 2024 report) | 2024 | 84.8 | 15.2 | 75.8 (white, grey, black) | 2.50 | 0.83 | 5.7 | 8 |
| Global (BASF 2024 report) | 2024 | 81.0 | 19.0 | 73.0 (white, black, grey) | 2.66 | 0.77 | 6.3 | 11 |
| Global (Axalta 2025 report) | 2025 | 81.0 | 19.0 | 74.0 (white, black, grey) | 2.68 | 0.81 | 6.4 | 10 |
| Global (BASF 2025 report) | 2025 | 83.0 | 17.0 | 75.0 (white, black, grey) | 2.59 | 0.78 | 6.0 | 10 |
| Measure | Data it needs | Rises when | Blind to |
|---|---|---|---|
| Achromatic share | Shares of named colours | More items are filed under white, black, grey or silver | How the neutral total is split; where the source draws the line between grey and a colour |
| Top-three concentration | Shares of named colours | The three leading colours gain, neutral or not | Everything below third place; merged categories |
| Normalised entropy | A full distribution | Shares become more even | Which colours are similar; changes in the number of categories |
| Effective colours, exp(H) | A full distribution | Shares become more even or categories are added | The contents of an 'other' bucket |
| Mean and median chroma | A colour value per item | Items are more saturated | Hue; the gap between a nominal value and a real surface |
| Muted or near-neutral share | A colour value per item | More items fall under the chroma threshold | Anything about items just above the threshold |
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
Colourwise analysisModerate evidence
In UK new-car registrations the combined share of the three leading colours rises from 62.5% in 2022 to 65.8% in 2025 while the share of grey, black, white and silver together changes by 1.4 points.
Based on: Sums of the market shares printed in SMMT's top-ten tables for 2022 and 2025: the three largest colours of each year, and the four neutral colours.
Caveat: Two years of a four-year series; the 2022 shares are printed to one decimal place.
Source: Colours put in the shade as grey matters most to UK car buyers; Grey in the fast lane as green powers up in Britain's top new car colours
Colourwise analysisStrong evidence
The effective number of colours in Germany's registrations, calculated as the exponential of Shannon entropy over seven categories, is 5.4 in 2014 and 5.2 in 2023.
Based on: exp(−Σ p ln p) over the six published colour groups and the remainder, each as a share of all new passenger cars registered in the year.
Caveat: The remainder is treated as one category although it holds several colours, which understates the true spread.
Source: Neuzulassungen von Personenkraftwagen in den Jahren 2014 bis 2023 nach ausgewählten Farben
FactStrong evidence
Shannon entropy of a set of shares is zero when one category holds everything and reaches its maximum, the logarithm of the number of categories, when all categories are equal.
Colourwise interpretationModerate evidence
Measures of named-colour shares and measures of chroma describe different things and should not be converted into each other or shown on one scale.
Based on: Colourwise's own method: categorical sources record a colour word with no colour value, and a threshold applied to measured values to imitate a category is itself the finding.
Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.