What a brand identity agency actually delivers, deliverable by deliverable

Search "brand identity agency" and Google hands you two kinds of page. Agency homepages, and listicles of agencies. Almost nobody publishes the thing a founder actually needs before signing: a list of what arrives in the shared drive at the end, and how to tell whether that list is complete.

This is the Brand Appart blog, so treat what follows as informed rather than neutral. We run brand identity projects for funded startups in Paris, which also means we read a lot of competing proposals, usually because a founder forwards three and asks which one is real. The pattern is consistent. Proposals differ less on price than on scope, and the scope difference hides in the last third of the document, where the application work either exists or quietly does not.

A useful comparison of studios by stage and specialism already exists in our guide to the best startup branding agencies for funded companies. This piece answers a narrower question. What is inside the engagement itself.

Positioning and verbal identity

Every identity project should open with words, not shapes. The output is a written positioning statement, a defined audience, a competitive frame, a messaging hierarchy, and a tone of voice with real examples rather than adjectives.

Good looks like a document you can hand to a new salesperson on day one and have them describe the company correctly. It names the category you compete in, says who you are not for, and gives three or four message pillars with sample sentences for a landing page, a cold email, and an error state in the product. Naming work sits here when it applies, including a clearance check against existing marks. The USPTO trademark search is the free first pass in the US, though not a substitute for counsel.

The common shortfall is a mood board with words on it. If the strategy phase produces "bold, human, innovative" and nothing that could be falsified, it has produced nothing. Another version of the shortfall: a beautiful forty-slide strategy deck that never gets referenced again because no single page of it is operational. Ask for the one-page version.

Logo and mark system

Not a logo. A system. The deliverable is a primary lockup, a secondary or stacked lockup for constrained space, a standalone symbol or monogram, an app icon drawn separately rather than cropped from the wordmark, and a favicon tested at 16 pixels.

Good looks like files that survive their worst use case. The mark works in one colour, reversed out of a photograph, embroidered, and shrunk to the corner of a slide. Clear space and minimum sizes are specified. Vector files are properly outlined, with correct file formats for each context, not a folder of PNGs at one arbitrary width.

The common shortfall is one lockup in one orientation, delivered as an SVG and a JPEG, with no app icon. Six weeks later the team is stretching the horizontal logo into a square avatar. A related shortfall is the mark that only works at presentation scale, because it was reviewed exclusively on a large screen in a dark keynote template.

Type and colour system

Colour and type are where an identity either becomes usable or becomes a maintenance problem. The deliverable is a type scale with named roles, licensed typefaces with the licence tier matched to actual usage, a primary palette, an extended palette for interface states, and documented contrast behaviour.

Good looks like a palette that passes accessibility at the sizes you actually use. The WCAG contrast minimum is the reference point, and a serious studio will have tested body text, disabled states, and the brand colour on both light and dark surfaces before shipping. Good also means the font licence covers web, app, and any desktop seats your team needs, with the cost stated in the proposal rather than discovered later.

The common shortfall is a palette of five beautiful colours with no functional greys, no semantic colours for success, warning, and error, and a brand colour that fails contrast the moment it becomes a button label. Type gets the same treatment: two display weights chosen for the deck, nothing specified for dense tables or small print, and a typeface licensed for print only.

Motion

Motion is the most commonly skipped deliverable and the one that most reliably separates a system from a logo. At minimum you want logo animation for intros and loaders, defined easing curves, duration ranges, and a stated point of view on how the brand moves.

Good looks like motion principles expressed as reusable values that a developer can implement, delivered alongside a short reference video. If the brand is meant to feel precise, the durations are short and the easing is sharp, and that decision is written down.

The common shortfall is a single logo animation delivered as an MP4 with no source file and no principles. It plays at the start of one video, then never again, and the product team invents its own motion language from scratch.

Brand guidelines

The guidelines document is the artefact most people imagine when they think about a brand identity agency engagement. It is also the one most likely to be padded.

Good looks like a document written for the people who will use it. Marketing needs templates and rules for social and campaign assets. Engineering needs values, not adjectives. New hires need the short version. Practically, that means a concise core document, correct and specific, plus a live Figma library where the components actually live. Guidelines that exist only as a static PDF go stale within a quarter.

The common shortfall is a hundred-page PDF full of aspirational photography, three pages of actual rules, and no editable source. The subtler version: guidelines that describe the identity without governing it, so no rule covers what happens when marketing needs a colour that does not exist yet.

Application to product UI and sales collateral

This is where proposals diverge most. An identity that stops at the brand layer leaves the product looking like a different company, which for a software business is most of the customer experience. We wrote about why that gap is expensive in our guide to B2B SaaS branding.

Good looks like the identity applied to real surfaces. Key product screens redesigned in the new system, not mocked on a dribbble-style poster. A website design or at least a homepage and one template. An investor deck and a sales one-pager, since those are the artefacts that actually circulate. Email signatures, social templates, and slide masters that someone can edit without a designer.

The common shortfall is a set of application mockups rendered as flat images. A phone in a stylised hand, a billboard on a Parisian street, a tote bag. These are portfolio shots, not deliverables. The test is simple: can your team open the file and change the text.

Design tokens and handoff

The last deliverable is the one that determines whether the identity survives contact with your roadmap. Tokens are the machine-readable version of the visual decisions: colour, spacing, type, radius, and motion values named once and consumed by both design and code.

Good looks like a token set exported in a standard format, a Figma library with published components and variables, a documented naming convention, and a live handoff session with your engineers rather than a link dropped in Slack. The Design Tokens Community Group format is the emerging standard worth asking for by name. We have written about how we keep that layer in sync across Figma, Claude, and GitHub.

The common shortfall is a Figma file with no published library, hex codes hardcoded into a spec document, and no session with the people who will implement it. Six months later the product has four near-identical blues and nobody can say which one is correct.

The deliverables at a glance

How to read a proposal and spot a logo-only engagement

Proposals are written to look comprehensive. A few reliable tells cut through that.

Count the nouns in the deliverables list, not the phases. A proposal with five impressive-sounding phases and four actual output files is a logo project with a strategy preamble. Scope lives in the noun list.

Look for the word "presented". Deliverables described as presentations or explorations are meetings, not files. "Three logo directions presented" tells you what happens in week four. It does not tell you what you own in week ten.

Check whether application work is quantified. "Brand applications" with no number attached will shrink to whatever fits the remaining budget. "Homepage plus two internal templates, investor deck to twenty slides, four product screens" is a commitment.

Find the handoff. If the proposal never mentions a working session with your engineers, a component library, or file ownership, assume none of those exist. Ask directly who owns the source files and under what licence, and whether the Figma library will be published to your organisation or shared as a view-only link.

Notice what happens after delivery. An identity with no support window gets misapplied within weeks, because the first ten edge cases always arrive after the project closes. Some studios include a support period, some offer an ongoing design subscription, some sell a second project. All three are acceptable. Silence is not.

Finally, compare the case studies to the deliverables list. If a studio's portfolio is composed entirely of poster-style compositions and no product screens, that is an accurate preview of what you will receive, regardless of what the scope section promises.

Take one proposal you are holding, strike out every line that is a meeting rather than a file, and see what remains. If the remainder is a logo, a palette, and a PDF, you are buying a mark, not an identity. That may be the right purchase at pre-seed. It is the wrong one if you have a product, a sales team, and a roadmap, because each of those will need the system you did not buy. If you want a second read on a scope you have been sent, start a project conversation or send it to a studio you are not considering hiring.