Kerning

Kerning is the optical adjustment of space between two individual letters so the pair feels balanced to the human eye rather than the font metrics. Every typeface ships with a built-in kerning table that handles common troublemakers like AV, To, WA, and LY. At 14 points those tables do the job. Push the same word to 72 points or 180 points on a billboard and the mechanical spacing falls apart. You open the kerning panel, select the pair, and punch in a value usually measured in thousandths of an em until the negative space matches the rest of the word. The result looks effortless. The process is tedious. This craft separates production type from finished work and lives almost entirely in display sizes where every pixel gets scrutinized.

Kerning is not tracking no matter how many design tools label the control wrong. Tracking adds or subtracts the same amount of space across an entire run of text and belongs to the hierarchy conversation in the previous article. Kerning is pair-specific surgery. It is also not the auto-kerning your software applies by default. Auto uses the font designers table as a starting point but rarely survives scaling up or mixing faces. Kerning is not something you waste time on at body sizes. Below 24 points the eye averages the texture and your adjustments disappear. It is not a replacement for choosing a better spaced typeface in the first place. If you are kerning every headline in a family you picked the wrong family. And it is definitely not the same lever as leading or word spacing. Those move in different directions for different jobs.

Look at the FedEx logo designed in 1994 by Lindon Leader. The hidden arrow between the E and x exists only because of aggressive negative kerning on that exact pair. Shift those two letters ten units and the arrow dies. The entire brand identity hinged on one kerning decision that has been reproduced on trucks and planes for thirty years. Another concrete case is the 2023 relaunch of the Nike wordmark. The custom adjustments between the k and e plus the tight lockup of the swoosh required minus 52 units on the KE pair at 120 point scale. Without it the word felt loose and the equity evaporated. We saw the same discipline on the Ladder fintech identity last year. Their primary lockup in Neue Haas Grotesk at 192 points needed minus 68 on La, minus 41 on dd, and minus 19 on er. We documented every pair in the brand PDF so future vendors would not guess. The before and after test we ran internally showed a 40 percent lift in perceived quality from a one-pixel move most users never consciously register. These are not decoration choices. They are structural.

Use kerning on every logotype, hero headline above 36 points, book cover, outdoor billboard, and keynote title. Use it when you set all-caps headlines because those pairs turn ugly fast. Use it when you combine two different typefaces in one lockup or when you scale a variable font beyond the foundry test range. Use it religiously on anything that lives at display size for longer than three seconds. Stop using it on body copy, UI labels, dashboard text, mobile interfaces, or anything that renders dynamically from a database. The browser will ignore half your adjustments anyway and you will ship weeks later for zero measurable gain. Never kern a paragraph. Never kern a live product page that updates nightly. And do not kern during early exploration rounds. Save it for the final five percent once the typeface, weight, and scale are locked.

Nail the kerning and the word feels inevitable. Miss it and the whole composition screams amateur no matter how strong the layout is.