Colour and what it is actually for. A working guide for small brands

Colour is the part of a brand people talk about most and decide on least. In almost every first conversation someone says the new colour should feel fresh, modern, premium, serious, young. Five words that mean everything and nothing. What usually comes out the other end is a blue that looks like the blue of the three nearest competitors, and nobody remembers why.

This piece is an attempt to do it differently. It covers what colour actually does inside a brand, what it cannot do, how to build a palette that survives a few years of daily use, and which numbers in all of this are not up for debate. It got long because the subject is long. If you only want the short version, read the headings and you are done in four minutes.

What colour actually does

Colour has three jobs inside a brand, and only three. It makes things recognisable, it sorts things, and it sets a tone. Everything else you hear about colour either belongs to one of those three or is wishful thinking.

Recognition is the strongest of the three. A surface in a particular colour is identified faster than a mark and far faster than a name. On a shelf, in a feed, on a poster wall, the colour impression decides whether anyone looks at all. That is why large brands hold on to their colour for decades even while they redraw the mark twice. Colour is the last thing you change.

Sorting is the least glamorous job and the most useful one. Colour says what belongs together and what does not. A price sheet with four offers becomes readable the moment one of them sits on a different surface. Instructions get understood faster when every step carries its own marker. Nobody notices this work. It only shows up when it is missing.

Setting a tone is the job people discuss most and judge worst. Colour does have an effect, but it is weaker, more culturally variable and far more dependent on context than most guides claim. There is a section on that further down.

What colour cannot do: it will not rescue a layout whose structure is wrong. It will not make an overlong text readable. It does not replace a promise. Paint a weak brand in bright colours and you have a bright weak brand.

The test that takes ten minutes

Before we talk about palettes, run a test you can do yourself. Take your current design, whether that is a website, a flyer or a deck, and remove every colour. Every tool has a way to do this. A greyscale screenshot will do at a pinch.

Look at the greyscale version and answer three questions. First: where does the eye go first. Second: is it clear what belongs together. Third: is the most important sentence still the most prominent one.

If any of those stays unclear, you do not have a colour problem. You have a problem with size, spacing, weight or order. Colour will not cure that, it will only cover it up. This is why we build every design in black and white first. Not out of severity, but because a structure that holds in grey holds even better in colour.

There is a side effect. You immediately see whether your design works for people who perceive colour differently. More on that below.

How many colours a brand needs

Fewer than you think. For a small or medium business four values are usually plenty, and even that is generous.

• A base tone for surfaces and text. Usually a very dark grey rather than pure black, because pure black looks harsh on screens and closes up in print.
• A paper tone. White or a very light grey, the ground everything sits on.
• One accent. The single colour that makes the brand yours, doing the work on buttons, markers and highlights.
• One or two signal colours for states. Something for errors, something for success. These are not brand colours, they are tools.

Anything beyond that needs a reason. A company with six product lines that have to be told apart on a shelf has that reason. A trade business with one service does not.

The usual cause of an oversized palette is uncertainty. When nobody can decide, everything gets kept. The result rarely reads as rich. It reads as arbitrary. A brand with one colour held consistently sticks in the mind better than one with seven that keep moving around.

How a palette gets built

We work in four steps, and the first one has nothing to do with colour.

Step one is the question of what you want to stand apart from. We collect the visual identities of the five nearest competitors and lay them side by side. If four of them are blue, then blue is the most expensive colour in that field. Not because it is bad, but because you would be spending money to look like everyone else. This overview takes half a day and prevents most of the bad choices.

Step two is the base tone. Not the accent, the base. Does your brand live on paper white or on deep black. That single decision shapes the impression more than any accent does. A site on black reads differently, costs differently in print and makes different demands of every photograph you will ever use.

Step three is the accent. Here we are not looking for the prettiest colour but for one that meets three conditions. It has to hold enough contrast against the base. It has to be producible in print without disappointing anyone. And it must not already be occupied in your competitive field.

Step four is the scale. From the accent we derive two or three lighter and darker values for backgrounds, lines and hover states. One thing matters here: these values are calculated, not guessed. We mix in the Oklab colour space rather than in RGB, because a step towards lighter looks even there instead of suddenly turning grey and muddy.

What comes out is a table of eight to twelve values, each with a note on what it is for. Not: light blue. Instead: surface behind notices, on paper white only.

Contrast is a number, not an opinion

This is the section that saves the most money in practice, because it ends the longest arguments.

Whether two colours are readable together is not a matter of taste. There is a formula. It compares the relative luminance of both colours and returns a ratio somewhere between 1 and 21. Black on white gives 21. Two identical colours give 1.

The accessibility guidelines name two thresholds. Normal text needs a ratio of at least 4.5. Large text, meaning roughly 24 pixels and up, or 19 pixels and up when bold, needs at least 3. Controls and graphics that carry information also need at least 3.

In Germany these numbers have been binding for public bodies for years, and since June 2025 the accessibility act covers many private providers selling to consumers online as well. Anyone running a shop, a booking system or banking services is inside that scope. Micro businesses under ten staff and under two million euros in annual turnover are exempt for services, which does not make the rule any less sensible.

In practice this means a colour pairing gets measured before it enters the guidelines. That takes thirty seconds. A light grey on white that looked elegant in the design regularly fails, and not by a small margin. Discovering that after the print run means paying twice.

One thing many people miss: the number applies to the colour as it finally appears. Text at fifty per cent opacity on white is not the text colour, it is the blend of the two. What gets measured is what the eye sees.

Colour blindness is more common than you think

Roughly eight in a hundred men and about one in two hundred women of European descent perceive colour differently from the majority. Most often that affects telling red from green. In a company with fifty men, four people on average see your red status marker and your green one as the same colour.

One rule follows from that, and it appears in every colour system we write: colour must never be the only carrier of information. A red outline on a form field is not enough, there has to be a sentence too. A green line in a chart is not enough, it needs a label or a different stroke. A required field is marked by the word required, not by a coloured star.

This rule costs nothing and makes designs better for everyone else too. A chart with labels on the lines reads faster than one with a legend off to the side, no matter how well someone tells red from green.

You can check this with any colour vision simulator. Ten minutes per design and the common traps are found.

Why the same blue looks different in print

A screen emits light. Paper does not. That is the whole reason, and it has large consequences.

On a screen colour is made of light coming out of the device. On paper it is made by ink absorbing some light and the rest bouncing back. A glowing neon green that shines on screen simply does not exist on uncoated paper. It is not hard to hit. It is not there.

Anyone planning printed work should know four things.

• Four colour process printing from cyan, magenta, yellow and black covers a smaller range than a good screen does. Vivid oranges, violets and greens suffer most.
• A spot colour, mixed to a fan guide, hits the tone exactly and adds cost to every print job. For a brand that prints a lot it pays for itself quickly.
• Paper changes everything. The same colour looks rich on coated stock and dull on uncoated, because the fibre drinks the ink. Choosing the paper early saves a disappointment.
• A page from the office printer proves nothing. If you want certainty, ask the printer for a proof, meaning a binding colour sample, before the run.

Our guidelines therefore carry more than a screen value for every colour. They carry the process mix and, where one exists, the spot colour, plus a sentence saying which of them wins in case of doubt.

What is true about colour psychology

The claim that blue builds trust and red makes you hungry turns up in every second guide. The truth is less convenient and more interesting.

First: effects of colour are measurable, but small. Context, shape, wording and prior experience override them easily. The same colour behaves differently on a package than on a government form.

Second: much of it is learned rather than innate, and it varies across cultures. White reads as purity across western Europe and as mourning in parts of east Asia. If you operate internationally, know that before you claim a meaning.

Third: the strongest effect is not the meaning of a colour but its fit. A colour that matches what the brand promises works. One that argues with it gets in the way. An organic business in toxic neon pink has a harder time, not because pink is bad, but because the expectation goes unmet.

And fourth: meeting the expectations of an industry is comfortable and expensive. Look like everyone and you get mistaken for everyone. The smart route often sits in between. Meet the expectation in the base tone and deliberately miss it in the accent.

Colour on screens and in motion

Two things work differently on screens than on paper, and both get forgotten.

The first is dark mode. A lot of people keep their device dark all the time, and a brand designed only for white looks poor there. A colour that reads strong on white often reads garish on black. The answer is not a second palette but one palette that brings its own steps for both base tones. This is exactly why we write colours down as roles rather than fixed values. Foreground, background, muted, line. What those roles mean changes with the section.

The second is calibration. No two screens show colour the same way. Whoever decides on colour should do it on a device set up for the job, not on a phone in direct sunlight. For print approval the rule is simpler still: the decision happens on paper.

Writing colour down so it survives

A palette that lives only in the designer head is lost the moment that person leaves. So every colour system comes with a document answering three questions per colour.

• Which values apply. For screen, for process print, for the spot colour, and if you build software, as a named variable too.
• What the colour is for. One sentence will do, but it has to draw a line. Buttons only, never body text.
• What it may be combined with. Best written as a small table with the measured contrast values, so nobody has to guess.

Alongside that comes a list of what is not allowed. No gradients between two brand colours. No text in the accent on a coloured ground. No coloured shadows. Prohibitions like these sound severe and are the reason a brand still holds together two years later.

If you run software or a website, store the colours as variables in exactly one place. Then a change is one line rather than a search through three hundred files.

The seven most common mistakes

• Too many colours, because nobody wanted to decide. You can spot it when none of the colours has a clear job.
• The accent used as a text colour. Almost every accent is too light or too saturated for longer text, and the measurement says so immediately.
• Colour as the only carrier of information. Red means error, but only for the people who see red.
• Values guessed rather than calculated. Lightening by twenty per cent in RGB often tips towards grey and looks dirty.
• The competitor colour. Show up in the same blue as the market leader and you are paying for their advertising.
• No plan for black. Pure black is harsh on screens and closes up in print. A very dark grey is nearly always the better call.
• Colour last. Think about colour only at the end and you will use it as a rescue. It only works as a tool.

A process that works

If you want to take this on yourself, here is the order we use on paid projects too. For a small company it runs to about two days.

• Collect the competition and lay the colours side by side. Note which tones are taken.
• Decide the base tone. Paper or black. Everything else hangs off that.
• Pick three candidates for the accent and show each of them in real use rather than as a square. A button, a heading, a surface, an envelope.
• Measure every pairing and drop the failures before taste enters the room.
• Calculate the steps and check them once in greyscale.
• Run the colour vision simulation.
• Write it all down, with jobs, values and prohibitions.

What this process leaves out is the question of which colour you like best. That is deliberate. It comes last and decides between the candidates that passed everything else.

What a colour system costs

For a small company the palette is part of the branding rather than a separate invoice. If you only need the colours because the mark and the type are already settled, we are in the low hundreds, because the work is contained. The expensive part is not the choosing. It is the checking and the writing down.

What really costs money is fixing it afterwards. A colour that disappoints in print costs a print run. A colour that misses the contrast threshold costs a rework of the entire application. Both are avoidable, and both happen regularly because colour gets treated as a matter of taste.

Common questions

Can I protect a colour legally. Within very narrow limits, yes. An abstract colour mark can be registered in Germany and in the European Union, but only with proven acquired distinctiveness, meaning a large share of the relevant public already links that colour to your company. For a young business that is not a route. Use the colour consistently and you build the claim over years.

Is one value enough for screen and print. No. The screen value is the starting point. The print mix is derived from it and usually adjusted by hand. Hand over a single value and you leave the translation to the printer.

How often should colours change. As rarely as possible. Colour is the slowest part of a brand. If it stops fitting after two years, the selection was usually the problem rather than the passage of time.

Is black and white boring. It is the opposite. A brand without colour has to work through shape, size and typography, which is considerably harder than a coloured surface. Get black and white right and colour becomes easy. And a single accent in an otherwise quiet setting hits harder than seven colours side by side.

What about gradients. They are rarely the problem and rarely the solution. A gradient between two very close values can give a surface some life. A gradient between two brand colours usually ruins both, because a tone nobody chose appears in the middle.

Colour names matter

What you call your colours sounds like a footnote and is not one. A name like light blue describes the appearance and says nothing about the purpose. Two weeks later a second light blue turns up in a file because somebody needed it slightly lighter, and a year later there are five.

Names built around the job stop that. Base, foreground, muted, line, surface, accent, warning. Name a colour after its job and you can swap the value without making the name wrong. A team that talks about jobs also argues less about tones.

For brands with several product lines a second level helps. A short code for the line, then the job. It reads clumsily and it prevents exactly the moment when somebody uses one line colour in the other line brochure because it happened to be at hand.

One more thing. Only give names to values that are genuinely in the palette. Every colour that gets a name eventually gets used. Hand out twelve names and twelve colours will end up on the page, even if four were meant.

Colour in tables and charts

The moment numbers are involved, colour changes role. It stops being decoration and becomes part of the statement, so a mistake is no longer a blemish. It is wrong information.

Three rules cover almost everything. First: order needs a sequence. If your data runs from low to high, the colour has to do the same, so light to dark within one tone rather than a trip through the rainbow. Second: categories need distance. When nothing in the data implies an order, the colours need clearly different lightness so they stay distinguishable in greyscale. Third: nobody can map more than six categories. Group them or label directly on the line.

The classic mistake is the traffic light. Red and green next to each other in a chart is one colour for a sizeable share of readers. If you want to show states, bring lightness into it or put a shape in front. An arrow up and an arrow down still work in the black and white copy being passed around the meeting room.

Tidying up an existing palette

Most companies that come to us do not need a new palette. They need order in the old one. The process is unglamorous and takes a day.

Start by collecting everything actually in circulation. Website, brochure, forms, templates, signage, the deck from sales. Pull out the colour values and put them side by side in one list. There are usually somewhere between fifteen and forty values where three were intended.

Then consolidate. Anything separated by less than a visible step becomes one value. After that, assign a job to each remaining colour. Whatever fails to get one is dropped. That single question, what job does this colour have, usually cuts the list to a third.

Finally measure the remaining pairings and write the palette down. If the effort seems steep: the expensive part of an untidy palette is not the mess. It is the time every future template costs because somebody has to decide all over again.

To finish

Colour is a tool with three jobs and a set of rules you can measure. The part everyone discusses, the question of feeling, is the smallest one. The part that decides whether it works is consistency. The same colour in the same place, for years, without exceptions.

People sometimes ask whether we have a favourite colour, and the honest answer is paper white. Not because we shy away from colour, but because a quiet surface is the reason an accent registers at all. A page where everything is coloured has no accent left. It just has colour. The difference between those two is the difference between a brand and a leaflet.

If you are in the middle of this and unsure whether your palette holds, run the greyscale test and measure your three most common pairings. That costs half an hour and answers most of the questions. Then leave it a week and look again. Colours that thrill on day one are not always the ones you want to live with for three years.

And if you would rather hand it over: this is exactly the work we do, for small and medium businesses in and around Stuttgart, at a fixed price, ending with a palette that is written down and measured. A first conversation costs nothing and takes twenty minutes. Send us a paragraph about what you have in mind and we will tell you honestly whether we are the right people for it.