Graphic design

What brand identity is, and how it differs from a logo

Brand identity is the set of repeated decisions that make a business's separate pieces of work look like they came from one author: colour, type, tone, kind of imagery and the shape of its applications. A logo is one element inside that set and is not an identity on its own; the test is simple, cover the logo and see whether you are still recognisable.

  • Lesson 10 of 12
  • Intermediate
  • Free, no signup

An identity from the core outwards

Each outer layer takes its decisions from the one inside it. If the core is unwritten, the outer layers change every time.

The promise
  1. 1

    Voice and tone

    How you write. Formal or familiar, terse or explanatory; which one matters less than keeping it.

  2. 2

    Colour and type

    One primary colour, one accent, the rest neutral. A Persian font with a clear licence.

  3. 3

    Mark and imagery

    Photography or illustration, plain or busy. The layer that usually goes unwritten.

  4. 4

    Applications

    Site, Instagram, invoice, packaging, sign. The only layer anybody sees.

The customer only ever sees the outer layer. The three inner ones exist so that it comes out the same every time.

Last checked: Facts and tool names in this lesson are re-checked against their sources on this date.

Brand identity is the set of repeated decisions that make a business's different pieces of work look like one author's. A logo is one of those elements, usually the smallest by surface area and the loudest by the amount of argument it causes.

The diagnostic is one line: cover the logo. If the Instagram post, the invoice and the site's front page still resemble each other and still point at you, you have an identity. If nothing but the logo connects those three, you have a logo and no identity.

That difference is not abstract; it costs money. A business that only bought a logo decides everything from scratch every time: what colour is this post, what font is this banner, what light is this product shot. Three months later it has ten pieces of work that are not siblings, and the customer remembers nothing, because nothing was repeated long enough to be remembered.

The reverse holds too: a good identity makes the work cheaper. Once colour, type and tone are decided, making a new post is a five-minute task rather than a meeting.

The layers of an identity, from the core outwards

The core is not visual at all. The core is what you promise and to whom. "Genuine spare parts with a replacement guarantee" is a promise; "superior quality at a reasonable price" is not, because nobody claims the opposite. Without that sentence the other layers are built on air, and they change every time somebody's taste does.

The next layer is voice and tone, meaning how you write. Formal or familiar, terse or explanatory, with humour or without. For the Iranian market there is a practical note: the official-formal register feels safe, but it makes everybody sound the same, and that resemblance is itself a decision you took without meaning to.

The third layer is colour and type, and those two do most of the work. One primary colour, one accent, everything else neutral; for most small businesses that is enough. Choose a Persian font with a clear licence. Vazirmatn is one example whose licence its authors publish themselves, the SIL Open Font License version 1.1, which means you can use it on the site and in a print file without worrying per project.

The fourth layer is marks and the kind of imagery: real photography or illustration, plain backgrounds or busy ones, people in the frame or not. This layer usually goes unwritten, and it is then exactly where the work falls apart.

The outer layer is applications: the site, Instagram, invoices and letterhead, packaging, the shop sign. It is the only layer a customer ever actually sees. The three inner layers exist so that this one comes out the same way every time.

The order of purchase for a small business

The order matters. Swapping the first station for the second is the most expensive mistake on this route.

  1. 1

    The name and its availability

    Domain, handles and the trademark situation; before any spending.

  2. 2

    A mark that works at both extremes

    At site-icon size, and on a sign from across the street.

  3. 3

    Two colours and one font

    A Persian font with a clear licence, not one of unknown origin.

  4. 4

    Three ready templates

    Post, story, and invoice or letterhead. The three you need weekly.

  5. 5

    One page of rules

    Codes, sizes and what not to do. It gets read, unlike a brandbook.

This route is for a small business. A company with ten people producing work at once needs a fuller document.

What does a small Iranian business need first?

Not a brandbook. A sixty-page brandbook is for a company with ten people producing work for it at the same time. For a small business the list of what is actually needed has five items, in this order.

First, the name and its availability. Before a single rial goes on design, check that the domain and the social handles are free, and ask about the trademark situation for that name. Changing the name after ten thousand cards are printed is the most expensive mistake on this list. Trademark registration has its own legal process and we do not advise on it; we only say to go and check before the design rather than after it.

Second, a mark that works at two extremes: at 16 pixels next to a browser tab, and on a shop sign seen from across the street. A mark that is only beautiful at large sizes loses half of its applications.

Third, two colours and one Persian font with a clear licence. Fourth, three ready templates: a post, a story, and an invoice or letterhead. Fifth, a one-page sheet listing the codes, the sizes, and the things that must not be done.

That one-page sheet beats any brandbook, for a reason everybody knows and nobody writes down: it gets read. The person making you a banner tomorrow will not open a sixty-page PDF; they will open one page.

How do you know the identity is working?

Four tests, each a few minutes long, each giving a clear answer.

  • The covered-logo test. Remove the logo from three different pieces of work. If they still belong together, you have an identity.
  • The single-colour test. Print the mark in solid black. If it dies without its gradient and colour, you will have trouble on stamps, invoices and any one-colour printing.
  • The sixteen-pixel test. Look at the mark at site-icon size. If it turns into a smudge, you need a simplified version.
  • The stranger test. Put two of your posts among ten others and ask somebody who does not know your brand which two came from the same place.

There is a fifth test, harder and worth more: give a freelancer who has never worked with you nothing but that one-page sheet and ask for a banner. If what comes back resembles your work, your identity is written down. If they had to ask questions, those questions are exactly what the sheet is missing.

And a limit worth stating plainly: consistency is not quality. An identity that is consistent and bad is still bad. These tests tell you whether your system holds together; they do not tell you whether your choices were good.

An identity, or just a logo

Identity audit sheetThe test: does it point at you without the logo?

Repeatable, therefore an identity

  • The colour codes are written down and identical everywhere
  • The exact font name and its weights are known
  • The mark has a single-colour and a small version
  • A one-page rule sheet exists that a stranger can follow

One-off, therefore only a logo

  • Each post's colour is picked from that post's photo
  • Each banner's font is whatever the app had handy
  • Only a PNG of the logo exists
  • Every new designer has to ask from scratch

The second column is not wrong, it is just not repeatable. Every row of it is a decision taken again from scratch each time.

Why does everyone buy a logo and almost nobody build an identity?

Because a logo is one defined item with one defined price, and an identity is a decision you then have to keep. But the size of that gap is measurable, and we publish it on this very site.

The demand table on our graphic design page shows that in Iran "logo design" gets about 2,900 searches a month against about 390 for "graphic design", which is seven times as many. The money side runs the other way: the advertising cost per click is $4.46 for graphic design and $1.01 for logo design. Everybody wants a logo; the person searching for graphic design has a budget.

For you as a business owner those two rows mean something simple: the market you stand in is full of people who buy only a logo, and that makes work which merely looks consistent more distinctive than you would expect.

The practical route is not expensive either. Buy the logo, but ask for three more things in the same order: the colour codes, the exact font names, and that one page of rules. If your designer does not hand those over, you have still only bought a picture.

We list our visual identity packages on the graphic design services page, where you can also see what gets delivered; but if you are working with another designer, ask them for that same list.

The fast path, with AI

The fast path here is not asking a model to build your brand identity. An identity comes from the reality of the business and the model does not have that reality. What a model does do well is this: turn your raw answers into a one-page brief that lists decisions as decisions, and, more valuable, write the rejection rules, meaning how anyone can tell that a piece of work does not belong to this brand. That clause is what turns a brief from a nice text into a working tool. A cheap fast model is enough; our current pick is in the <a class="text-link" href="/en/ai/">AI section</a>.

  1. Write six short answers: what you sell, to whom, what you promise that a competitor does not, the word a customer uses for you today, where you are seen, and who makes the work.
  2. Run the prompt below with those six answers. Note that it explicitly forbids the model from choosing colours and fonts; that is deliberate, because those two are visual decisions that need a look at the market and the competitors.
  3. Judge the output by one rule: delete every sentence a competitor could write about themselves word for word. What survives is your real brief, even if it is short.
  4. Test the rejection rules against real work. Take the last banner you made and see whether those rules reject it or accept it. If the rules reject nothing, they are not rules.

Copy-ready recipe

Role: someone who writes brand identity briefs. You have only the answers
below and know nothing else about this business.

Answers:
1. What we sell: {...}
2. To whom: {...}
3. The promise a competitor does not make: {...}
4. The word a customer uses today: {...}
5. Where we are seen: {...}
6. Who makes the work: {...}

Write a one-page brief with these sections:

a) The brand promise in one sentence, from answer 3 only; if answer 3 is
   something every competitor also says, say exactly that and invent no
   sentence.
b) Tone: three adjectives, each with one short example sentence in that tone.
c) Three things this brand must not resemble, each with a one-line reason.
d) Rejection rules: five signs by which anyone can say a piece of work does
   not belong to this brand. Every rule must be observable, not a matter of
   taste.
e) The list of applications in order of importance, from answer 5 only.

Rules:
- Do not choose colours or fonts and do not suggest any. Those are visual
  decisions and they do not follow from these answers.
- Write no claim about the market or competitors that is not in the answers.
- Delete any sentence a competitor could write about themselves verbatim.
- If an answer is empty or vague, list it as a question at the end.

Before you trust the output: A brief is not a decision; it is the minutes of your decisions. Two things the model cannot do and you must: look at what competitors produce, to see where you have started resembling them, and choose colour and type, both of which need eyes on the work. And if a brand has not launched and its name and strategy are not public, remember that sending these answers to an outside service is a decision.

AI in this kind of work

In identity work AI takes over two specific jobs and does not take over a third. What it does well: organising a business owner's scattered statements into a brief, and producing visual sketches so the argument stops being verbal. What it does not do: decide. Decisions need knowledge of the market and the customer, and the model has seen neither your market nor your customer.

Tools that actually help

  • Claude Suits the brief and the rejection rules, because it paraphrases less when you ask it to keep your own sentences and adds fewer marketing lines. Iran is on neither of Anthropic's two supported-countries lists.
  • Gemini Suits condensing a large pile of customer comments and messages and pulling out the words customers themselves use. Our own entry notes Iran is not on Google's supported-countries list; the payment route is in <a class="text-link" href="/en/ai/buy/">buying AI access</a>.
  • RGB AI image ranking For visual sketching and mood boards, the ranked table of image models with criteria and sources is here. Their output is raw material: it is not a vector file and it says nothing about whether a mark is unique.

Where it backfires

The first risk in this subject is one nobody treats as a risk: tonal convergence. If you write your brand's copy on a model's default settings you will end up with a tone that thousands of other businesses also have, which is the exact opposite of what an identity exists for. The fix is simple and belongs in the brief itself: set the tone yourself with three adjectives and a few example sentences, and judge generated copy against them. The second risk is legal. A mark a model produced is, first, not a vector file and has to be redrawn; second, its commercial use depends on that service's terms; and third, more important than either, the model does not know whether the mark it made resembles somebody else's registered trademark. For something going on a sign, an invoice and packaging, those three together mean the model's output belongs at the sketch stage rather than at handover. The third risk is about data: the strategy of an unlaunched brand and a name that has not been announced are confidential business information; sending them to an outside service is a decision, and whether the conversation is used to train a model depends on that service's plan and settings and is written on its own page.

Sources: Anthropic: is my data used for model training Google: Gemini Apps Privacy Hub

Where this advice stops

A visual identity does not create demand. If nobody wants the product or the price is wrong for the market, the most consistent identity in the world will not sell it; at best it makes the same number of people remember you sooner. This lesson is also about the visual system rather than naming and trademark law; that is work for a lawyer and a registration specialist and we do not advise in that area. And one thing seen often in the Iranian market: an identity designed for Instagram alone usually fails in print and on a sign, because neither the gradient nor the on-screen colour survives out there.

From our own work

That "everybody buys a logo and almost nobody builds an identity" is not a guess; we publish it on this site and you can go and look. Our graphic design page carries a demand table whose Iranian rows come from Google Ads through DataForSEO with no country filter, because that service has no country filter for Persian. In that table "logo design" runs 2,900 searches a month against 390 for "graphic design", while the advertising cost per click runs the other way, $1.01 for the logo term and $4.46 for the graphic design term. Those two rows carry the whole difference between a purchase item and a decision: the thing everybody wants is cheap, and the thing only the funded go looking for is expensive. The keyword difficulty and search intent columns are empty in that table's Persian rows, and that is deliberate too: neither is available for a Persian phrase and we do not fill a gap with an estimate.

Real follow-up questions

Do I need a brandbook to start?

No. One page with the colour codes, the font names and a few plain rules works better for a small business than a sixty-page document, because it actually gets read. A brandbook becomes necessary when several people are producing work for you at the same time.

How many colours does a brand need?

One primary colour, one accent, and everything else neutral; for most small businesses that is enough. Every extra colour is an extra decision that has to be made again on every new piece, and that is where consistency goes.

Should I make my logo with AI?

For sketching and finding a direction, yes, and it is useful. For the final mark, no, for three reasons: the output is not vector, commercial use depends on that service's terms, and the model does not know whether the mark it produced resembles somebody else's registered trademark.

Keep learning