There should be no generic phrases like "the font must be readable" in the guideline — that does not work as an instruction for a web designer or email marketer. Comprehensive brand books, such as those of Yandex Practicum or the British Museum, are built around a table: font family, weight, point size, and tracking for each heading level (H1–H6), body text, and captions — separately for web, print, and email.
| Element | Point size | Line spacing | Tracking |
|---|---|---|---|
| H1, Montserrat ExtraBold | 32–48px | 1.2–1.3 | -30…-40 |
| H2–H3 | 24–28px | 1.25 | less than 25 (usually -10…-20) |
| Body text | 14–16px | 1.3–1.5 (for 14px — 21pt) | 0…-10 |
| Captions, caption | 11–12px | 1.3 | 0 |
Note: for large headings, negative tracking is used with bold weight — letters visually "merge" if zero tracking is left, especially with geometric grotesques like Montserrat ExtraBold. For small text, negative tracking kills readability — only small negative values or zero are acceptable here.
This is the most underrated part of typography in the brand book. Gmail supports only its own built-in fonts — Roboto and Google Sans — meaning that half the audience will see a fallback font instead of the brand font. If it is not set intentionally, Outlook on Windows silently reverts to Times New Roman, and the message will look as if it was designed in 2005.
The solution is to explicitly define the alternative through a fallback font stack in CSS font-family (for example, font-family: Montserrat, Arial, sans-serif), rather than relying on the email client to pick something similar. The brand book should include a ready-made chain: brand font → web-safe alternative with similar proportions → system font as the final resort. For example, for Montserrat, a logical chain would be Montserrat → Arial → sans-serif, not Montserrat → Times New Roman by default.
A good example of this approach is to use a simplified version of the font for email newsletters, supported by the specific email layout system, while keeping logos and banners inside the email in the corporate font as images. This is not a compromise, but a conscious division: where the font is rendered as text, use the simplified version; where it is an image, use the original.
Approving a font pair solely based on a Figma mockup is a mistake even experienced teams make. Before finalizing the typefaces in the final document, it is necessary to perform a mandatory set of checks:
The rule of ‘no more than two or three fonts’ is valid for most brands, but not universal. A third and fourth font are justified in specific situations, not by the designer’s whim:
Brand unity in these cases is maintained not by the number of typefaces, but by the strictness of the rules: each additional font must have a clearly defined area of application and must not overlap in context with the main two.
A brand guide without a section on 'what to do if the main font is unavailable' is a brand guide that will break on the first non-standard medium. Below is the minimum set of redefinitions that should be explicitly written out channel by channel, rather than left to the improvisation of the designer.
| Channel | Main font | Redefined |
|---|---|---|
| Website | Branding font via WOFF2 | System sans-serif until font loads (font-display: swap) |
| Branding font in OTF/TTF | No — license must cover print usage separately | |
| Simplified web-safe analog | System font via mso-font-alt | |
| Social Media | Brand font only in images/carousels | Platform font for native post text |
| Internal documents | Simplified font (Calibri, Arial) | Not required — document is not a public medium |
A separate point is licensing. A web font license almost never includes the right to embed it in a PDF for printing or in an email as text. If the brandbook does not specify the license status for each channel, the designer will learn about the restriction at the moment when the print shop has already taken the order into work — and this is not a hypothetical risk, but a frequent cause of urgent layout corrections just a day before printing.
A good brandbook differs from a bad one not by the number of pages about fonts, but by the fact that it eliminates all questions for the person who is typesetting a letter on Monday morning without access to the designer. The minimum working team for the section:
Without these five points, the brand book remains a beautiful PDF that looks good on presentations, but doesn't prevent an email in Outlook from opening in Times New Roman. With them, the brand typography works consistently predictable on the website, in the newsletter, and on the metro banner, regardless of who is laying out the design and in which program.