Both are legitimate. Neither is automatically the grown-up choice. The skill is diagnosing which problem you actually have before you spend anything, because the failure mode is not doing too much or too little in the abstract, it is doing the wrong one and having to do the other anyway.

What each one actually is

A refresh takes a brand that is fundamentally sound and sharpens it. The logo is redrawn rather than replaced. The colour palette is tightened, the typography updated, the photography and tone brought into line. The name stays. The core idea stays. What changes is the quality and consistency of how it all shows up. A good refresh makes a brand look like it has its act together, without a customer ever quite being able to say what moved.

A rebrand changes the substance. A new name, or a new position in the market, or a fundamentally different idea of who the brand is for. The visual identity follows from that, but the visual identity is not the point. The point is that the answer to "what is this company" is now different, and everything has to be rebuilt to tell the new story.

The tell is simple. If the strategy is right and only the execution is letting you down, you need a refresh. If the strategy itself no longer fits the business, no amount of execution will save it, and you need a rebrand.

A refresh keeps the brand and improves how it shows up. A rebrand changes what the brand is.

When a refresh is the right call

The business has not fundamentally changed, but the brand has aged. It was made in-house, or a decade ago, or by three different hands who never spoke, and it now looks tired next to competitors rather than wrong.

The identity is inconsistent rather than misconceived. The logo is fine but there are four versions of it, the colours drift, and the website, the deck and the signage look like three different companies. Most of what people mistake for a weak brand is really a brand with no system holding it together, and that is a refresh, not a rebrand.

You have real equity you would be mad to throw away. People recognise the name, the colour, the mark. Recognition is expensive to build and cheap to keep, so unless it is actively working against you, keep it and make it better. This is the same discipline as a brand <a href="/notes/what-a-brand-rollout-involves/">rollout</a>: the value is in consistency applied everywhere, not novelty.

When you actually need a rebrand

The business has changed and the brand is now describing a company that no longer exists. You have moved upmarket, or changed what you sell, or merged, or outgrown the name you chose when it was just you at a kitchen table. The brand is not tired, it is inaccurate, and no amount of polish fixes an inaccurate brand.

You are entering a new market where the existing brand does not translate or does not carry. This is common in travel and hospitality especially, where a name that works in one territory can mean nothing, or the wrong thing, in another. When we helped DerTour establish itself in the UK the work was not cosmetic, it was building brand foundations for a market that did not yet know the name.

The brand is actively holding the business back. It signals the wrong price, attracts the wrong enquiries, or embarrasses the people who have to hand out the card. That is a strategic problem wearing a visual costume, and refreshing the visuals leaves the problem exactly where it was.

No amount of polish fixes an inaccurate brand. That is the one thing a refresh cannot do.

The expensive mistake, in both directions

Refreshing when you needed a rebrand is the common one, because it is cheaper and less frightening. You spend a modest sum, the brand looks a little better for a season, and the underlying mismatch is untouched. Within a year or two you are back, having paid for a refresh and now paying for the rebrand you needed all along. The refresh was not wrong, it was just answering a question you were not really asking.

Rebranding when a refresh would have done is rarer but not unheard of, usually driven by a new hire who wants a visible win or a board that mistakes change for progress. You throw away hard-won recognition, confuse loyal customers, and spend a large sum solving a problem you did not have. The classic warning sign is a rebrand with no strategic reason behind it that anyone can state in a sentence.

Both mistakes come from starting with the deliverable instead of the diagnosis. Decide what is actually wrong first. The right project falls out of that answer on its own.

A test you can run in five minutes

Write down, in one sentence, what your business is and who it is for today. Then look at your brand and ask whether it is telling that story. If the story is right and the telling is scrappy, you need a refresh. If the story itself has moved on, you need a rebrand.

Then a second question, because the first can flatter you. Ask what has actually changed since the brand was made. If the honest answer is "the market moved and we look dated", that is a refresh. If it is "we are a different company now", that is a rebrand, however much you would prefer it not to be.

If you cannot tell, that is useful information too. It usually means the strategy is unclear, and clarifying it is the first stage of either project. That is where the discovery work earns its place, and it is the same first step whichever way the answer lands.

What each one costs

Broadly, a refresh costs less than a rebrand because it is doing less: no naming, less strategy, and it builds on foundations that already exist rather than laying new ones. A rebrand costs more and takes longer because the thinking is deeper and the number of things that have to change is larger.

But the ranges overlap more than you would expect, because a thorough refresh of a large business can involve more application work than a lean rebrand of a small one. Rather than repeat the figures here, we set out the honest numbers, and what moves them, in <a href="/notes/what-a-brand-identity-costs/">what a brand identity actually costs</a>. The short version: the cost is driven by scope and application, not by which word you use for the project.

Why we are telling you this

Because talking you into the bigger project would be easy and we would rather you spent the right money once. Half of doing this well is refusing to sell someone a rebrand they do not need, and the other half is being honest when a refresh will not fix a problem that is really about strategy.

If you are not sure which one you are looking at, that is the most useful conversation to have before anyone quotes anything. Tell us what has changed about the business and we will tell you, plainly, which project it points to, even when the answer is the smaller one. You can see how we think about the work across <a href="/sectors/travel/">travel</a> and <a href="/sectors/hospitality/">hospitality</a>, or just <a href="/contact/">start a conversation</a>.

Neither is much help when you are trying to work out whether you can afford to do this properly, and whether doing it properly is worth it.

The numbers

Brand identity work starts at £5,000. That covers positioning, a visual identity and guidelines for a smaller business: enough to be consistent, enough to hand to a printer or a developer without them guessing.

Most identity projects for an established SME land between £10,000 and £25,000. That range buys proper discovery, naming if you need it, a system rather than a logo, and enough application work that you can actually use the thing on the Monday after we finish.

Above £25,000 you are usually paying for scale rather than better design. More stakeholders to bring along, more markets, more touchpoints, research that has to stand up in a boardroom, or a rollout that runs for months after launch.

Below £5,000 we would tell you to spend it elsewhere. Not because cheap design is beneath us, but because under that number nobody can afford to do the thinking, and identity without the thinking is decoration you will replace in two years.

Below £5,000, nobody can afford to do the thinking. Identity without the thinking is decoration you will replace in two years.

What actually moves the number

Four things, roughly in order of how much they matter.

<strong>How many people have to agree.</strong> This is the big one and almost nobody expects it. A founder who can decide in the room is a different project to a board of six with a marketing committee. Not because consensus is bad, but because every additional decision-maker adds rounds, and rounds are the thing you are paying for.

<strong>Whether it needs a name.</strong> Naming is its own discipline and it is slower than people think. Generating candidates is quick. Finding one that survives a Companies House search, a trademark screen, a domain check and being said out loud in a meeting is the work. Add £3,000 to £8,000 depending on how contested the category is.

<strong>How many surfaces it has to work on.</strong> A consultancy that needs a website, a deck and a document template is a smaller job than a hospitality brand that needs signage, menus, uniforms, packaging, vehicle livery and a booking journey. The identity is the same size; the system around it is not.

<strong>Whether you want us to stay for the rollout.</strong> Handing over a PDF is cheaper than being there while a hundred people apply it. It is also how most good identities quietly decay.

What you are paying for that isn't the logo

The logo is the smallest part of the job and the part everyone fixates on. If you priced our time by output, you would conclude we charge several thousand pounds for a shape.

What you are actually buying is a set of decisions that hold. Who you compete with and what they all sound like. What you can credibly say that they cannot. The typeface, the colour, the layout logic, the tone, and crucially the rules for what happens when the system meets something nobody anticipated. A trade stand. A spreadsheet. A shop window in bad weather.

You are also buying the ability to defend it. When a well-meaning colleague asks why the logo isn't bigger, someone needs an answer better than personal taste. Guidelines that explain the reasoning survive that conversation. Guidelines that only show the artwork do not.

How we structure it

Every project runs in four stages: discover, define, design, deliver. The cost is split evenly across them, and we only move to the next stage when you are happy with the last.

That structure exists to remove the single biggest anxiety a business has about commissioning design, which is handing over a lot of money at the start and hoping. You are never more than a quarter of the budget away from a decision point.

You get a fixed price in writing before anything starts. We have never sold hours, partly because measuring design in hours rewards the wrong things, and partly because nobody wants an invoice that arrives as a surprise.

You are never more than a quarter of the budget away from a decision point.

What cheap actually costs

There is a version of this where you spend £500 and get something usable. It happens. If you are testing an idea and expect to rebrand within eighteen months, that might be the right call, and we would say so.

The failure mode is subtler than bad design. It is design that works in exactly one context and falls apart everywhere else. A logo that only reads at one size. A colour that dies in print. No type rules, so eleven people pick eleven fonts. Six months later you are paying someone to sort it out, and you have spent the difference twice.

The other cost is comparative. Cheap design tends to be trained on what already exists, so it produces the average of your category. If the point of the exercise was to look unlike your competitors, average is the one outcome you cannot use.

A better first question

Cost is a fair thing to ask and we are not going to be coy about it. But it is rarely the question that decides whether this works.

The more useful one is: what has to be true in twelve months for this to have been worth doing? More inbound enquiries. Charging more without losing people. Being taken seriously by a client you cannot currently reach. Recruiting people who currently pick your competitor.

Answer that and the budget usually answers itself, because you can see what the outcome is worth. It also tells us what we are actually solving, which changes what we make.

If you want to talk it through, we will give you a straight answer on scope and cost, including whether we think you need us at all.

What happens next decides whether any of it worked. We do a lot of this part, including for Kuoni, whose brand we did not create but were brought in to carry into retail, campaign, print and digital. It is less glamorous than a rebrand and considerably more consequential.

What rollout actually is

Rollout is the work of getting a new identity onto everything the business touches, and making sure it survives the journey.

That sounds procedural. It isn't. A brand presented in a deck exists in perfect conditions: chosen photography, generous space, one designer controlling every pixel. Rollout is the opposite. Fixed dimensions somebody else specified. Print processes that flatten anything subtle. A colleague in a regional office updating a template at half past four on a Friday.

The identity has to hold in those conditions, and the only way to know whether it does is to take it there.

A brand presented in a deck exists in perfect conditions. Rollout is the opposite.

The unglamorous inventory

The first real task is finding out how many things there are. This is almost always a shock.

A mid-sized business will typically have somewhere between two hundred and a thousand branded items in circulation. Email signatures. Invoice templates. The PowerPoint everyone in sales has quietly customised. Vehicle livery. Signage with a lead time measured in months. Packaging with print runs already committed. Trade stand graphics in a warehouse. A recruitment ad running until March. Social profile images across six platforms, two of which nobody can remember the password for.

Nobody has this list. Building it is the first fortnight of most rollouts, and it is the point where the scale of the job becomes real for everyone involved.

Where brands actually come apart

Three places, reliably.

<strong>Print.</strong> Screens are forgiving and paper is not. Colours shift, thin type fills in, a gradient that looked considered turns muddy. Anything that only exists as a screen mockup has not been tested.

<strong>Retail and physical space.</strong> A high street window is the least forgiving test there is. Mixed lighting, fixed dimensions, viewed at speed from an angle by someone who was not looking for you. Signage also costs real money and takes real time, so mistakes here are expensive in a way a bad web page is not.

<strong>Templates other people use.</strong> This is the quiet killer. If the document template is awkward, people will work around it, and within a year you have eleven versions of your brand and no way back. Templates need to be easier to use correctly than incorrectly, which is a design problem rather than a documentation one.

Rules for the cases nobody planned for

Good guidelines are not a catalogue of approved artwork. They are a set of principles someone can reason from when they hit a situation the guidelines do not cover, which they will, constantly.

The test we use is whether a competent person who has never met us could make a sensible decision from the document alone. Not the right decision necessarily. A defensible one.

That means writing down why, not just what. Why the logo has that much space around it. Why the secondary colour is used sparingly. A rule with a reason attached survives a stakeholder who disagrees. A rule without one gets overridden by whoever is most senior in the meeting.

How long, and who should do it

For an SME, expect two to four months of active rollout after the identity is signed off, with a long tail as things come up for renewal. For a business with retail estate or committed print, twelve to eighteen months is normal.

There are three ways to run it. The studio who made the identity stays and does the work. Your in-house team does it with the studio supporting and reviewing. Or you hand over and hope.

The third one is the cheapest and the most expensive. We have been called in to fix the results more than once, and repairing an inconsistent rollout costs more than doing it properly would have.

The middle option is usually the right answer for anyone with an internal team. They know the business, we know where the identity is fragile, and the reviewing is the part that matters.

Handing over and hoping is the cheapest option and the most expensive one.

What good looks like

The measure of a rollout is not whether the launch looked impressive. It is whether the brand still looks like itself in eighteen months, in the hands of people who were never told why any of it was decided.

If it does, someone did the unglamorous work. If it doesn't, the identity was probably fine and the rollout was left to chance.

That is why we would rather be judged on the second year than the launch. It is also why, when a client asks what happens after the reveal, the honest answer is that the reveal is roughly the halfway point.

Both worries are reasonable and both rest on the same misunderstanding, which is that design is mostly production. If design were mostly production, faster production would mean better design, and everyone would already be watching that happen. It is not what is happening.

What actually got faster

Plenty, and we are not precious about it. The tools are genuinely good at the work that sits either side of the thinking.

Research and synthesis. Before a brand gathering we want to arrive knowing the sector, the competitors, the language everybody in that market has quietly agreed to use. That used to be two days of reading. It is now a morning, and the time saved goes into better questions rather than a smaller invoice.

Naming, at the volume stage. When we <a href="/work/formara/">named Formara</a> we needed hundreds of candidates before we had five worth defending. Generating volume is a task, not a craft. Knowing which five survive contact with a Companies House search, a trademark check and being said out loud in a meeting is the craft, and that part did not move.

Build. Front-end scaffolding, boilerplate, the tedious middle of a component library. <a href="/services/digital-product/">Digital product work</a> is where the honest efficiency gains are largest, and they are large enough to change what a project of a given size can include.

Rollout and adaptation. On a job like <a href="/work/kuoni/">Kuoni</a>, where a brand has to hold steady across retail, campaign, print and digital, a great deal of the labour is repetition under constraint. Resizing, re-cropping, versioning, checking. The tools take a real bite out of that, and nobody involved mourns it.

The tools are very good at the work either side of the thinking. They have not touched the thinking.

What did not move at all

The brief. Not the document, the actual brief, which is the thing you only find by sitting with people and noticing the gap between what they say the business is and what it is. It is the first half of <a href="/services/brand-identity/">every brand identity project</a> and nobody writes it down for you, because if they could, they would not need us.

In a gathering the useful moment is almost always accidental. Someone contradicts a colleague. Someone describes a customer in a way that does not match the deck. A founder gets animated about a detail everybody else considers minor, and that detail turns out to be the whole proposition. There is no prompt for that. You have to be in the room and paying attention.

Judgement under ambiguity is the second thing that did not move. A model will give you a hundred options and no opinion. Choosing is the job. Choosing means being able to say why this one and not that one, holding that position when a client pushes, and being right often enough that people keep hiring you. Confidence about which option is correct is not something you can generate.

And taste, which is an unfashionable word for accumulated judgement. It comes from having made things, watched them fail, and understood why. It is why <a href="/about/">the &amp;Co in our name</a> is a group of specialists rather than a set of subscriptions.

The homogenisation problem

There is a commercial risk in this that we think is underrated.

These tools are trained on what already exists, which means their instinct is always the average of the category. Ask for a <a href="/sectors/travel/">travel brand</a> and you will get the travel brand you have already seen, competently. Ask for a consultancy and you will get the reassuring blue thing.

That is fine when you want a competent version of a known quantity. It is a serious problem when the entire point of the exercise is to look unlike the market you are in. If every studio reaches for the same tools trained on the same body of work, the market converges, and the businesses that converge with it lose the only thing brand was supposed to buy them.

Differentiation was always the scarce asset. It is scarcer now, and it is getting more valuable at the same rate that production is getting cheaper.

Ask a model for a travel brand and you get the travel brand you have already seen. Competently.

The line we hold

So the rule is simple enough to say in one sentence. We will use these tools anywhere they save time on work that is not the thinking, and nowhere near the part where the decision gets made.

In practice that means a few specific commitments. No client work is presented that a person at this studio cannot defend line by line. No generated imagery goes into a brand system without being made properly afterwards. Nothing confidential goes into a tool we do not control. And no efficiency gain is allowed to quietly become a reduction in how much thinking a project receives, which is the failure mode nobody notices until the work is average.

That last one matters most. The obvious way to use these tools is to do the same work for less money. The better way is to do more of the work that only people can do, on the same budget, because the rest of it stopped taking so long.

Why we are telling you this

Partly because you were going to ask. Mostly because the answer says something useful about how we charge.

We have never sold hours, and this is why that turns out to have been the right call. If you are buying production, then yes, the price of production is falling and it should keep falling. If you are buying judgement, the arrival of an infinite supply of competent options makes it worth more, not less. Someone still has to know which one is right.

That is the whole position. The machine is fast, and speed was never the part we were short of.

If you would rather see this applied than argued, the <a href="/work/">case studies</a> are where it shows. And <a href="/field/">The Field</a> is the same instincts pointed at coffee and watches, which is a lower stakes way of watching us think.

At the concentrations found in coffee leaves it's genuinely lethal to insects. It disrupts the nervous system, causes paralysis, and kills. The plant isn't offering a stimulant. It's running a chemical weapons programme, and it has been doing so for considerably longer than we've been brewing it.

Then it found a better use for the weapon

Here is where it becomes interesting, and where the story stops being about poison and starts being about design.

Coffee and citrus plants also put caffeine in their nectar. Nectar is the opposite of a defence: it's the bribe, the thing the plant offers pollinators in exchange for carrying its genetic material to the next flower. Putting a toxin in the bribe seems like a category error.

In 2013 a team at Newcastle University, led by Geraldine Wright, worked out what was actually happening. Honeybees rewarded with caffeinated nectar were three times more likely to remember a floral scent than bees rewarded with sugar alone. The plant isn't poisoning the bee. It's drugging it into loyalty.

The plant isn't poisoning the bee. It's drugging it into loyalty.

The dose is the whole trick

The detail that makes this remarkable is the concentration. Caffeine in nectar never exceeds the bee's bitter taste threshold. Go over it and the bee tastes the compound, registers it as a toxin and avoids the flower. Go under it and there's no pharmacological effect at all.

The plant has to land in the gap between undetectable and ineffective, and it does. Natural selection calibrated a dose precisely enough to alter behaviour without ever being noticed.

There's a lesson in that for anyone whose job is influencing behaviour. The effective dose is almost always lower than the obvious one, and the moment your audience notices the technique, it stops working.

Which brings us to altitude

Specialty coffee sells altitude hard. Strictly High Grown, 1,800 metres, the thin air and the cool nights. The accepted story is that altitude slows the cherry's maturation, which concentrates sugars and produces a denser bean with more complex acidity. That part is well supported.

The caffeine question is messier, and the mess is the interesting bit.

A study of Arabica varieties in southwest Ethiopia found caffeine content falling as altitude rose, with the highest concentrations in lowland beans. The explanation follows directly from what caffeine is for: there are fewer insects at altitude, so the plant has less need to defend itself, so it invests less in the chemical defence. Caffeine as a cost the plant pays only when it has to.

Except a study of Arabica in Rwanda found the opposite, with caffeine increasing at altitude. Others have found effects in both directions depending on shade, variety and how the cherry was processed after picking.

So the honest answer is that altitude is one variable among several, and anyone telling you high-grown coffee is reliably lower or higher in caffeine is tidying up an argument that hasn't been settled. What altitude does reliably affect is flavour. That's worth paying for on its own.

The same molecule, in you

Caffeine works on the bee's memory by blocking adenosine receptors in the mushroom body, the part of the insect brain that handles learning and smell.

It does the same thing to you. Adenosine accumulates in your brain across a waking day and binds to receptors that make you feel progressively more tired. Caffeine is shaped closely enough to adenosine to occupy those receptors without activating them. It doesn't give you energy. It intercepts the message that you're running out.

This is why the crash arrives later. The adenosine hasn't gone anywhere. It has been queuing.

What the evidence actually supports

Coffee has accumulated a large body of observational research, and it's worth being careful about what it shows.

A meta-analysis of nine prospective cohort studies found higher caffeine intake associated with substantially lower Parkinson's risk, around 39% lower in men and 29% lower in women. The mechanism is the same adenosine antagonism, this time appearing to protect dopaminergic neurons. Notably, decaffeinated coffee doesn't show the effect, which points at caffeine specifically rather than coffee generally.

Coffee consumption is also inversely associated with type 2 diabetes risk across systematic reviews, with proposed mechanisms including antioxidant and anti-inflammatory effects and changes to the gut microbiome.

The caveat matters though: these are observational associations, not proof that coffee causes the benefit. People who drink a lot of coffee differ from people who don't in many other ways. Researchers themselves say it's premature to recommend drinking coffee as a means of preventing disease. Enjoy it because it's good. Treat the health headlines as encouraging rather than instructive.

Why any of this is on a design studio's website

Because it's the same problem we work on, solved by a plant.

The coffee plant had a compound that worked brilliantly at killing things, and found a second application where the identical molecule, at a fraction of the dose, changed behaviour instead. Not by being louder. By being precisely calibrated to sit just under the threshold of notice.

Most brand work fails in the other direction. Too much dose, too obvious a technique, an audience that spots the manipulation and disengages. The plant has been iterating for longer than we have and it arrived at restraint.

Sources

• Wright et al., Caffeine in floral nectar enhances a pollinator's memory of reward, Science (2013)
• Girma et al., Influence of altitude on caffeine, 5-caffeoylquinic acid and nicotinic acid contents of Arabica coffee varieties, Journal of Chemistry (2020)
• Rwandan Coffea arabica has caffeine increasing with altitude, ScienceDirect
• The effect of caffeine on the risk and progression of Parkinson's disease: a meta-analysis
• Coffee consumption and reduced risk of developing type 2 diabetes: a systematic review with meta-analysis
• Coffee, Linus Pauling Institute, Oregon State University

The usual explanation is cheap beans. That is true and it is not the whole story. The more useful thing to understand is that a supermarket bag and a small roaster's bag are solving completely different problems, and roast level is where you can see the difference most clearly.

Start with the date on the bag

This is the single most informative thing on any packet of coffee, and it is the thing almost nobody checks.

A good roaster prints a roast date. It tells you when the coffee was made, which is the only fact that lets you judge whether it is any good right now. Supermarket coffee prints a best before date, typically set six to twelve months after roasting, which tells you when the retailer stops being responsible for it.

Those are answers to different questions. One is about flavour and one is about liability. If a bag will not tell you when it was roasted, it is usually because the answer would not help sell it.

What is actually going wrong in there

Roasted coffee is chemically busy and it does not stay still. Two processes run at once.

The first is degassing. Roasting produces carbon dioxide inside the bean, which then escapes over the following days and weeks. This is why fresh bags have one-way valves, and why coffee roasted yesterday brews badly: the gas gets in the way of the water. Coffee needs a few days to settle before it is at its best.

The second is oxidation, and this one only goes one way. Oxygen degrades the lipids, phenolics and aromatic compounds created during roasting, producing the papery, flat, faintly rancid character of stale coffee. The Specialty Coffee Association's review of the staling literature notes that lipid oxidation into volatile compounds begins within the first twenty four hours of roasting and grinding, and that trained assessors could detect rancidity in coffee packed in air after four months.

The rate is strongly tied to how much oxygen the coffee meets. Work presented on the chemistry of coffee oxidation puts it at roughly a ten percent increase in degradation rate for every one percent increase in oxygen. Grinding is therefore the single most destructive thing you can do to coffee ahead of time, because it multiplies the exposed surface area enormously. Pre-ground coffee is stale by design.

A roast date is about flavour. A best before date is about liability. They are answers to different questions.

The roast spectrum, honestly

Roast level is a sliding scale of how much of the flavour comes from the bean and how much comes from the fire.

Light roasts stop shortly after first crack. Most of what you taste is the coffee itself: the origin, <a href="/field/coffee/caffeine-is-a-pesticide/">the altitude</a>, the variety, the way the cherry was processed. Acidity is high, the body is lighter, and the character is specific. It is also unforgiving, because there is nowhere for a defect to hide.

Medium roasts trade some of that specificity for sweetness and balance. Sugars caramelise further, acidity drops, body builds. This is where most people are happiest and where a lot of very good coffee lives.

Dark roasts push to or beyond second crack. Second crack is the point at which the bean's internal structure fractures and oils migrate to the surface. Origin character is largely gone by then. What you taste is roast: bitterness, smoke, dark chocolate, char. It is a legitimate style with a long tradition, and it is also extremely good at covering things up.

The thing almost everyone has backwards

Dark roast is widely believed to keep longer. It does the opposite.

Second crack physically ruptures the bean's cell structure, so a dark roasted bean is markedly more porous than a light one. Oxygen gets in faster. It has also driven oils onto the outside of the bean, where they are directly exposed to air and go rancid sooner. Dark roasts degas quicker and decline quicker. Light and medium roasts, particularly dense high grown beans, hold their peak for longer.

So why does the belief persist? Because dark roast does not last longer, it fails less noticeably. When the dominant flavour is already roast rather than origin, there is not much delicate character left to lose. The coffee was flattened at the roastery, so staling has less work to do.

That is a genuinely useful property if your coffee has to survive a warehouse, a lorry and four months on a shelf. It is a terrible property if you wanted to taste Ethiopia.

Which is exactly why the supermarket does it

Coffee is graded on a hundred point scale. Eighty and above is specialty. Below that is commodity grade, which is where nearly all mainstream supermarket coffee sits, often blended arabica and robusta from many origins at once. Robusta carries roughly twice the caffeine of arabica, which is <a href="/field/coffee/caffeine-is-a-pesticide/">not an accident of flavour but of defence</a>.

Lower grade coffee carries defects: sour, fermented, woody, musty notes from inconsistent picking and processing. Those defects are real and they are detectable, and there is a well known way to deal with them, which is to roast past the point where anyone can taste them. The char covers the fault.

Add the blending, which exists so the product tastes identical in March and September regardless of what the harvest did, and the shelf life requirement, and the pre-grinding, and you arrive at exactly the coffee that is on the shelf. Nothing about it is incompetent. Every decision is correct for the problem being solved. The problem being solved is distribution.

Dark roast does not last longer. It fails less noticeably, which is not the same thing.

What a small roaster is actually selling you

Three things, and only one of them is the coffee.

Freshness, structurally. A roaster working in small batches roasts to order or close to it, so the bag reaches you within days rather than months. That alone accounts for a large part of the difference in the cup, before you get anywhere near bean quality.

Information. A good bag tells you the farm or the co-operative, the altitude, the variety, the process and the roast date. This is not romance, it is traceability, and traceability is what makes it possible to pay someone more for doing something well.

And a different distribution of money. The 2026 Coffee Barometer estimates that farmers receive somewhere between three and six percent of the final retail price of coffee. Specialty green coffee trades at a substantial premium to the commodity C price, and direct trade relationships push further still, but the premium only reaches the farm if somebody in the chain is deliberately sending it there. A small roaster who visits origin and publishes what they paid is the mechanism by which that happens. The supermarket, by design, has no such mechanism.

Worth being honest about the limit of this argument: buying specialty coffee is not automatically an ethical act, and some research suggests certain specialty segments return less value to producing countries than commodity markets do. The thing that matters is the specific roaster and what they actually do, not the word on the bag.

The short practical version

Buy whole bean. Grinding at home is the largest single improvement available to most people, and it costs one purchase.

Buy from someone who prints a roast date, and drink it between roughly one and four weeks after that date.

Buy smaller bags more often rather than a kilo you will still be working through in August.

And if you have only ever had dark roast, buy one light or medium bag from a local roaster before deciding what you like. A lot of people who believe they dislike acidity have simply never had coffee that tasted of anything else.

Why any of this is on a design studio's website

Because dark roast is the most honest example we know of a decision that looks like a flavour choice and is actually an operations choice.

It happens constantly in <a href="/services/brand-identity/">brand work</a>. Something gets simplified because it photographs better in a deck. A name gets chosen because it was available rather than because it was right. A tone of voice flattens out because flat language survives being written by fifty different people in six offices. Every one of those is a reasonable answer to a real logistical problem, and every one of them costs the thing that made the brand specific in the first place.

The question worth asking, in coffee and everywhere else, is not whether a decision is defensible. It is which problem it was defending against.

We wrote about a related version of that trade in <a href="/notes/what-we-let-the-machine-do/">what we let the machine do</a>, which is about efficiency arriving in design the way it arrived in coffee.

Sources

• What is the Shelf Life of Roasted Coffee? A Literature Review on Coffee Staling, Specialty Coffee Association
• SCAA Member-Driven Research: Coffee Staling Report
• Mapping the chemistry of coffee oxidation: a new framework for freshness, World of Coffee
• Speciality vs commodity coffee, Union Hand-Roasted Coffee
• Inside the 2026 Coffee Barometer, Part 2: What Companies Are Not Saying, Daily Coffee News
• Understanding the Coffee C Market and its impact on green coffee prices
• Does packaging affect coffee freshness? MTPak Coffee

That growth curve is the story worth noticing. British watchmaking spent most of the twentieth century being politely described as heritage, which is usually what people say about an industry that has stopped. It hasn't stopped. It's rebuilding, and it's doing it through a lot of small companies rather than one revived giant.

The numbers behind the revival

The Alliance's second Bellwether Survey put some figures on what had been mostly a feeling. British horology has grown 65% since 2021, reaching £206 million in revenue, at a compound annual growth rate of 18%.

More than 75% of responding companies reported revenue growth across the last three trading years. Around 42% reported double-digit growth. Roughly 30% reported growth between 300% and 400%, which is the kind of number you normally only see in software.

Employment has moved too. The sector now supports about 1,600 jobs across full-time, part-time and freelance, which is a 3.6-fold increase since 2021.

For context, that makes it one of the fastest-growing specialist manufacturing sectors in the British Isles, achieved through a period of supply chain disruption, high energy costs and inflation. Most British manufacturing has not had that decade.

The category word problem

Here's the thing that should interest anyone who works in brand. Every single company in that hall is British. It's the entry requirement. So the one word each of them would most like to own is the one word none of them can.

This happens constantly and most businesses handle it badly. Organic. Artisan. Sustainable. Independent. The word that got you into the category cannot also be the thing that distinguishes you within it, and yet it's almost always the word that ends up largest on the stand.

The brands that work in a room like that have decided what they are beyond the obvious. A specific kind of engineering. A particular customer. A point of view about what a watch is for. Something that survives being surrounded by forty-seven companies making the same claim about provenance.

The word that got you into the category cannot also be the thing that distinguishes you within it.

Scarcity, done properly

Twenty-six special editions were made for the 2026 event, available only to people who turned up. Not a pre-order, not a mailing list drop. You had to be in the building.

Scarcity marketing is usually a bit grubby, because the scarcity is invented and everybody knows it. This version isn't. The constraint is real, the reason for it is obvious, and the reward goes to the people who made the most effort. That's a meaningful difference and it's why it doesn't feel like a tactic.

It also turns the event itself into the product, which is a considerably better position than being a stand at a trade show.

The queue is part of the design

A detail that says more about the organisers than any of the watches. For 2026 the day moved to a three-session structure, staggering entry to manage the crowds.

Popularity creating a worse experience is one of the most common ways a good thing quietly ruins itself, and the usual response is to do nothing and hope. Recognising it and redesigning the day around it is unglamorous, invisible when it works, and exactly the kind of decision that determines whether people come back.

Most brand problems are like this. Not the logo. The queue.

Small brands, real advantages

Watch companies of this size sit in an interesting position. They're too small to outspend anyone and too specific to be for everyone, which turns out to be a strength if you accept it rather than fight it.

They can make three hundred of something. They can answer their own emails. They can have an opinion that would be focus-grouped out of a larger business. Nearly all of the affection these companies attract comes from those constraints rather than despite them.

It's a familiar shape if you run a small studio. The temptation is always to present as bigger than you are. The opportunity is almost always the opposite.

Why this is in The Field

Because a mechanical watch has no practical justification. Your phone keeps better time, costs less and never needs servicing.

People buy them anyway, for how they were made, who made them and what owning one says. That's brand doing its actual job, in a category where the functional argument was lost decades ago and everyone involved knows it.

Which makes British Watchmakers' Day a useful place to look. Forty-eight companies, the same permission to use the same word, and a hall full of evidence about which ones found something else to say.

We work in this world

This isn't only a spectator interest. We work with the team behind three British brands in this category, all built out of Tewkesbury.

WatchGecko is the established one, grown from a side project buying and selling watches on eBay in 2012 into one of the better known British names in straps and accessories. Geckota is the in-house watch brand, and the name came from a family joke about watch geeking. ZuluDiver is the dive-focused sibling, built around tool straps for people who actually get them wet.

Three brands, one group, and a clear reason for each to exist separately. That's harder to get right than it sounds, and it's one of the more common things we get asked to fix.

Sources

• Alliance of British Watch & Clock Makers
• British watchmaking resurgence: sector grows 65% in just three years, Europa Star (Bellwether Survey)
• Global sales top £206 million for British watch businesses, WatchPro
• The British watch industry has grown 65% in 3 years, Time+Tide
• British Watchmakers' Day 2026 highlights, Oracle of Time
• British Watchmakers' Day 2026 review, MVS Watches
• British Watchmakers' Day showcases the strength of British horology, Mr Watchmaster