
Your Brand Guidelines Are a PDF. Everything Else in Your Company Is a System.

Francesco de Chirico
September 16, 2026
9
min read
Think about how much of your company already runs on systems.
Engineering has version control, so every decision has a history and a person attached to it. Product has a backlog. Finance has a ledger. Sales has a pipeline with stages that mean something. Even your front end has design tokens now, so a colour is a named decision rather than a hex code someone typed.
Then there is brand. A 60 page PDF, finished eighteen months ago, living in a Drive folder, opened about twice a year by people who interpret it slightly differently each time.
That was fine when the volume of brand output was low and the people producing it were few. Neither of those things is true any more.
What actually changed
Not the brand. The demand for it.
Five years ago, brand output came from a small group of people who had been in the room when the decisions were made. A designer, a writer, maybe an agency. The volume was naturally limited by how fast those people could work, and the quality was held in their heads rather than in the document.
Now everyone produces brand-adjacent work. Sales builds decks. Customer success writes help docs. Product ships empty states and error messages. Marketing runs six channels. And all of them now have tools that will generate something plausible in thirty seconds.
Production got cheap and distributed. Judgement did not. The result is a volume of output that nobody has the capacity to review, held to a standard that exists only in a document most of those people have never opened.
You can see the symptom in any growing company: a brand channel in Slack filling up with requests that are each individually reasonable, and collectively a sign that your best designers are spending their week reproducing decisions somebody already made.
Guidelines describe a brand. They cannot take part in it.
This is the structural problem, and it is not a quality problem. Plenty of beautiful, thorough brand guidelines fail at exactly this.
A guideline is a description. It sits outside the work and waits to be consulted correctly by someone who knows it exists, can find it, and will interpret it the same way as the last person. Every one of those is a point of failure, and they compound with every new hire and every new channel.
Worse, guidelines are written for the cases their authors imagined. They will tell you the logo clearspace and the campaign photography direction. Take a real question to them, what our personality does to an error message, which grey belongs in a data table, how formal the tone gets in a billing email, and most of them have nothing to say. Not because the authors were lazy, but because a document can only cover the situations somebody thought of in advance.
A phrasebook is not a grammar
Here is the distinction that matters.
A phrasebook gives you sentences. Useful, as long as the situation you are in is one of the situations in the book. The moment it is not, you are stuck, because a phrasebook cannot generate anything it does not already contain.
A grammar is different. A grammar gives you rules for forming sentences nobody has written yet, and the sentences still sound like the language. That is what you need from a brand once the output is being produced by fifty people and a handful of tools rather than three designers who were in the room.
Most brand guidelines are phrasebooks. They show you approved examples and hope the next situation resembles one of them.
This is also why consistency is the wrong goal. Consistency says everything looks the same. It is achievable, it is measurable, and it produces brands that are rigid in the places that matter and silent in the places that are not covered. Coherence is the better goal: a recognisable centre with room to vary. You want someone to be able to make something genuinely new and have it still feel like you. A phrasebook cannot do that. A grammar can.
What a brand system actually contains
If you have read our piece on what a design system actually contains, this will look familiar, because the logic is the same. Brand is further behind, that is all.
1. Decisions, written once, in one place
The point of view underneath everything: what the company believes, who it is for, how it behaves, what it will not do. Not a deck. A short, current, single source that a person or a model can be pointed at.
Most companies have this scattered across a strategy deck, a positioning doc, a tone of voice PDF and the founder's head. Scattered is the same as missing.
2. Tokens
The brand expressed as named values rather than described appearances. Colour, type, spacing, motion, named for purpose so they survive a rebrand. This is the layer that lets the brand actually reach the product, which is where most brands quietly die. We wrote about that in brand-led product design.
3. Patterns
How the brand behaves in a situation, not how it looks in isolation. What our voice does when we have to deliver bad news. How we write an empty state. What a sales deck opens with. The recurring compositions, which is the layer almost every brand skips and the one that answers real questions.
4. Governance
Who decides, how something gets added, what the review path is, how it stays current. Without this, a brand system is a snapshot that starts rotting the day it ships, which is precisely what happened to the PDF.
Brand book, brand portal, or brand system?
Three different things that get sold under the same name, so worth separating.
A brand book is the PDF. Still the right answer for a small team, a single market and low output volume. Cheap, fast, and genuinely sufficient at that scale.
A brand portal is the PDF on a website with asset downloads and version control. Better. Solves findability and staleness, which are real problems. Does not solve the structural one, because it is still a description that waits to be consulted. A portal is a well-organised phrasebook.
A brand system is the decisions, tokens, patterns and governance, built so they can be used by the people and the tools doing the work rather than read beforehand. The tokens flow into the product. The patterns answer the questions the book never anticipated. The decisions sit somewhere a model can be pointed at when it drafts.
Most companies need the second and are sold the first. A smaller number genuinely need the third, and they usually know it because they can feel the review load.
The part everyone underestimates
Building a demo of this is easy. Building one people use is hard, and the gap between them is where most of these projects die.
A system that gets it right 70% of the time can be worse than no system at all, because now every output needs checking and someone has to work out which 30% is wrong. That is more work than before, not less.
Which means the unglamorous requirements are the whole job. Everything it produces has to be editable rather than take-it-or-leave-it. It has to be honest about what it does not cover and route those cases to a human instead of guessing confidently. It has to be reviewable, so someone can see what it decided and why. And somebody's name has to be on keeping it current.
None of that is exciting. All of it determines whether the thing is alive in a year.
What stays human
Worth being precise, because this is where the argument usually gets lazy.
The system handles repeatable production: the fourth version of a deck template, the social crop, the help doc that needs to sound like you, the hundred small things that are decisions already made being re-made by hand.
Humans keep the brand-defining work: positioning, naming, the original visual and verbal language, and every judgement call about where the brand goes next. A system can apply a point of view. It cannot have one.
The teams getting this right are pairing two kinds of people rather than retraining one into the other. Someone who creates the original language, and someone who codifies it into something usable. Turning your brand designers into engineers is not the goal and mostly produces worse designers.
When you do not need any of this
Three situations where the PDF is correct and we would say so.
You are small and the output is low. Two or three people producing brand work, all of whom were in the room. The overhead of a system costs more than the inconsistency does.
Your brand is about to change. If you are heading into a repositioning, systematising the current brand encodes decisions you are about to throw away. Do the strategy first.
The real problem is that nobody agrees on the strategy. A system will faithfully distribute an unresolved point of view to every channel at once. If the leadership team cannot say what the company believes, no amount of tooling fixes that, and the tooling will make the disagreement more visible rather than less.
The honest test: you need a brand system when the cost of re-deciding things is being paid every week, by people whose job is not brand.
FAQ
What is the difference between brand guidelines and a brand system?
Guidelines describe the brand and wait to be read. A system encodes the decisions as usable assets, tokens, patterns and a governance process, so they reach the work without someone having to interpret a document first.
Do we still need a brand book?
Usually yes, as the human-readable front door. The point is that it stops being the only artefact. The book explains the thinking; the system does the work.
What is a brand portal?
A hosted, versioned home for guidelines and assets, with downloads and usually some access control. A real improvement on a PDF in a Drive folder, and still a reference rather than a participant.
How long does it take to build a brand system?
Foundations and tokens land in a few weeks. A working set of patterns with documentation and a governance process is a few months, depending on how many channels it has to cover. It is never finished, which is the point.
Who should own it?
One named person with authority to say no. Shared ownership reliably becomes no ownership, and this is the layer that rots fastest without an owner.
The takeaway
Your brand is the only part of the company still running on a document. Everything else got a system, because at volume, documents stop working and systems do not.
That is not an argument for more tooling. It is an argument for writing your decisions down once, in a form that can reach the work, and being clear about which parts are repeatable production and which parts are judgement you should never automate.
Most companies will use the new tools to produce more. A smaller number will use them to produce more of the same thing, on purpose, and that is a different outcome entirely.
If you are feeling the review load and wondering whether the answer is a better PDF or something else, that is the conversation to have. Start with brand guidelines, or a chat about design systems if the product side is where it is hurting most.

