Type does most of the work

Ask someone to describe a brand they use every day and they will describe a feeling, a colour, maybe a shape. What they are actually recalling, most of the time, is type — because ninety-something percent of what a company puts in front of them is words.

The mark appears once, the type appears everywhere

A logo sits in a corner. Typography carries the pricing page, the invoice, the onboarding email, the terms nobody reads and the error message everybody does. Every one of those is a brand impression, and every one of them is set in the same voice.

So when a budget puts weeks into the mark and an afternoon into the type system, it has funded the least-seen asset and rushed the most-seen one.

Choosing a face is a constraint problem

Taste comes last. Before anything is set in it, a typeface has to answer questions that have nothing to do with how it looks in a specimen at 200 points.

• What is the smallest size it has to survive? Many display faces collapse below 14px, and that is exactly where most of your words live.
• How many weights do you actually need? A family with four real weights beats one with eighteen you will never license consistently.
• What languages ship next year? Retrofitting an alphabet you did not plan for is a rebrand in everything but name.
• What does it cost at your traffic? Web licensing is priced on pageviews. Growth turns a cheap decision into a recurring one.

None of these is glamorous. All of them are the difference between a system that scales and one that gets quietly replaced by the nearest default.

A scale, not a list of sizes

Most type systems fail because they are a list — heading, subheading, body — rather than a scale with rules about when each step is used.

We define presets by role, not by pixel value: what a display line is for, what a lead paragraph is for, what a label is for, and what a piece of secondary metadata is for. The sizes behind them can change per breakpoint or per surface; the roles do not. That is what lets a team lay out a page nobody designed.

If two people on your team can pick different sizes for the same job and both be right, you have a list, not a scale.

Monospace, and other decisions with consequences

A monospaced face sets every glyph on the same advance. It reads as precise, technical, slightly editorial — and it runs about a third wider than a proportional face at the same size.

That is not a reason to avoid it. It is a reason to cut the scale for it: pull the tracking back hard at display sizes, drop caps above label size, and accept that your headline measure is shorter than it would otherwise be. A typeface swapped without re-cutting those numbers is a different brand wearing the old one’s proportions.

The cheapest improvement available

If you change one thing this quarter, change the measure. Set body copy to somewhere between forty-five and seventy-five characters a line, fix the line height at around 1.6, and leave everything else alone.

It is the least fashionable typographic advice there is, and it will do more for how your product reads than a new family ever could.