A brand is at its best on the day it is delivered.
The presentation goes well. Everyone agrees the mark is strong. The colours look expensive on the slide. Somebody says the word "elevated" out loud and nobody laughs. Then the deck gets filed, the PDF gets emailed to four people, and the brand starts its slow slide into whatever the intern picked from the colour wheel eighteen months later.
That is not a design failure. The design was fine. What failed is that nothing connected the brand to the place the brand actually lives, which is a content management system where people who never saw the deck publish pages every week.
So here is how to brand a business properly, in the order the steps actually have to happen, and what most people skip.
The logo is the smallest and least load-bearing part of a brand, and it gets ninety percent of the attention because it is the only part that fits on a slide.
A brand is a set of decisions that have already been made, so nobody has to make them again under pressure. What colour is a destructive button. How long can a headline get before it breaks. What does a photo of our team look like. What tone do we use when something has gone wrong. Which of these two words do we use, always, and which one do we never use.
Every one of those is a decision somebody will otherwise make at 4pm on a Thursday with no context. The value of a brand is the number of those decisions it has already settled.
Positioning first. Not because it is virtuous, but because a mark drawn before the argument is settled is decoration, and decoration gets redrawn the moment the argument changes.
You need four things written down in plain language before anyone opens a design tool:
Who it is for. Specifically. "B2B companies" is not an audience, it is a census category. "Marketing teams running HubSpot without a developer" is an audience, because you can picture one and you know what annoys them.
What you actually sell. Not the category. The thing the money buys. There is usually a gap between the two, and the gap is where most brands go generic.
What you are against. A brand with no opponent has no shape. It does not have to be a competitor. It can be a habit, a default, a way things are normally done badly.
The words. Ten to twenty of them. The ones you use, and the ones you have banned. This is the cheapest part of a brand to build and the part that survives longest, because words travel into emails, sales calls and support replies where no designer can follow.
If you cannot write those four things, the design stage will produce something that looks good and means nothing, and you will not be able to say why you dislike it.
Once the argument is settled, the visual work is mostly building a system that other people can use without you.
Colour, with the contrast checked. A palette is not a set of swatches, it is a set of pairs. Every combination someone might actually use — text on background, button on surface, link on tint — needs to pass contrast at the sizes you will really use, not at 48px. Most brands ship a beautiful accent colour that becomes illegible the first time it is used as body text, and then everyone quietly stops using it.
A type scale, not two fonts. Pick the sizes, the line heights and the weights, and pick how many there are. A scale with six steps is a system. A scale with nineteen steps is a mess with a font in it.
One spacing rhythm. Every gap in the whole brand comes from one small set of numbers. This is the single change that most reliably makes work look considered, and almost nobody does it.
An image direction that survives reality. Not a mood board of photography you will never be able to commission. A rule specific enough that somebody choosing a stock photo at short notice can tell whether a given image is allowed.
Here is the test that separates a brand from a presentation.
Before sign-off, build the pages nobody puts in the deck. The dense one with four hundred words and a table. The one with a single sentence on it and too much white space. The terms and conditions page. The 404. The form with three validation errors showing.
A brand that only works on the homepage hero is not finished, it is a poster. Most of the pages on a real website are the awkward ones, and if the system has nothing to say about them, whoever builds them will invent something, and the thing they invent becomes the brand by volume.
This is also the cheapest moment to find out that the accent colour is unusable on small text, or that the headline font falls apart at three lines. Finding it out after rollout costs a rollout.
This is the step that decides whether any of the previous work still exists in two years, and it is the step most branding projects treat as somebody else's problem.
The guidelines live in a PDF. The website lives in a CMS. Nothing connects the two. So every new page is somebody remembering the rules, or not, and the drift is invisible because it never happens in one visible jump. It happens one slightly-wrong padding value at a time until the site is forty shades of the same blue and nobody can say which one is correct.
The fix is to stop relying on memory. On HubSpot CMS, that means three specific things:
Every value becomes a token. Colour, type size, spacing, radius — each one is a theme setting feeding a CSS custom property, and nothing is hardcoded on a page or inside a module. When the brand changes, one setting changes and the whole site moves with it. When it does not, you are repainting three hundred pages by hand.
Module fields become choices, not free text. If an editor can type a hex code, an editor will type a hex code. Give them a choice list of the brand's four surface styles instead. The goal is not to restrict people, it is to make the on-brand option the easy one and the off-brand option genuinely difficult to reach by accident.
Wire up the account, not just the theme. HubSpot has a brand kit and a brand settings inheritance system that lets colours and logos flow from account settings into theme and module fields. Almost nobody sets it up. Do it, and the CRM emails, the meeting links and the forms stop being the three places your brand quietly does not apply.
When the work is done, there is one question worth asking, and it is not whether the site looks good.
It is: can somebody on this team build an off-brand page by accident?
If the honest answer is yes, the brand is a document, not a system, and its condition from here is a straight line downward. If the answer is no — if the wrong choice requires effort, or is not representable at all — then the brand will still be recognisable long after everyone who approved it has left.
That is the whole difference. Not taste. Enforcement.
Every rebrand quote you will ever receive is really a measurement of how well step five was done the first time.
If the site runs on tokens, a rebrand is mostly a settings change, staged and reviewed and switched over in an afternoon. If it does not, it is a page-by-page archaeology project, and the site sits half-rebranded for a month while somebody finds the last three templates with the old blue in them. HubSpot's own rebrand checklist is long for exactly this reason.
The same logic decides whether you can run a second brand on the same foundation without cloning your way into two codebases — which is a decision worth making deliberately rather than discovering.
Settle the argument, design a system rather than a picture, prove it on the pages nobody shows off, and then encode it where the pages are actually made — so being on brand is the default rather than an act of discipline.
That last part is the one that gets skipped, and it is the only one that determines whether any of the rest of it lasts.
We do all five. Branding and design for HubSpot covers the positioning, the identity and the visual system, and then the implementation into your portal as design tokens, theme settings, module defaults and email templates — so the site still looks like the brand a year after we have gone. From €6,000, quoted per project.
If the brand is already fine and it is the build that needs doing, that is a custom theme instead, and we will say so.