Most typography advice tells you what a font says. This tells you what it costs.

Every brief I've ever read wants the same four things from a typeface. Beautiful. Interesting. Trustworthy. Impossible to miss.

You can have some of that. You cannot have all of it, and the research on which combinations are available has been sitting in the Journal of Marketing since 2004.

The paper is Henderson, Giese and Cote, Impression Management Using Typeface Design. It's cited over 300 times, it's aimed explicitly at managers rather than designers, and almost nobody in brand strategy talks about it. This post is the version I wish I'd had eight years ago.

The chain: characteristics, dimensions, impressions

Most typography writing collapses everything into two steps: this font looks like X, therefore it means Y. The actual structure has three.

Layer one: letterform characteristics. The things you can literally draw or select. Curved or angular. Serifed or not. Slanted. Uniform. Ornate. Heavy. Condensed. Looks typed, or looks handwritten.

Layer two: six design dimensions. What those characteristics roll up into perceptually.

Layer three: four strategic impressions. What the viewer ends up feeling.

The reason this matters is that you work in layer one, the client briefs you in layer three, and almost every failed type decision happens because nobody mapped layer two in between.

The six dimensions

Henderson and colleagues had already established three of these in earlier work on logo design. The typeface study confirmed those three carried over, and added three that are specific to type.

The three carried over from logo research:

Natural. Loads positively on curved, organic, slanted and active. Loads negatively on "looks typed," in contrast to looks handwritten. This is the hand-versus-machine axis.

Harmony. Loads positively on uniform, balanced, smooth and symmetrical. How much the letterforms agree with each other.

Elaborate. Loads positively on distinctive, ornate, conveys meaning and depth. Loads negatively on readable and common use. Note that second half. Elaborateness is defined partly by being less readable and less familiar, which means the trade-off is baked into the dimension rather than being a side effect.

The three specific to type:

Flourish. Decorative embellishment. Swashes, expressive terminals, strokes doing more than they need to.

Weight. Stroke mass. Light through to heavy.

Compressed. Horizontal proportion. Condensed through to extended.

Those last three are the ones designers change constantly without treating it as a strategic decision. As you're about to see, weight in particular is doing far more than anyone gives it credit for.

The four impressions, and their exact recipes

This is the part worth printing out. Each impression has a specific formula, and they contradict each other.

Pleasing

High naturalness, high harmony, high flourish, moderately compressed, and NOT elaborate.

Pleasing is the easy one, and it's the one that gets confused with good. It's warm, liked, likely to survive a stakeholder review. It is also, structurally, the least distinctive option on the list, because being pleasing requires low elaborateness and elaborateness is where distinctiveness lives.

Engaging

High naturalness, high elaborateness, high compression, and LOW harmony.

Engaging requires disharmony. The letters have to disagree with each other. That's not a stylistic observation, it's a negative loading in the model.

And here's the cost. The paper's own factor analysis found the engaging factor loaded positively on interesting, emotional and innovative, and negatively on calm, formal, honest and familiar.

Read that again. The typographic profile that makes a brand feel innovative is the same profile that makes it feel less honest.

Reassuring

High harmony, high flourish, and LOW elaborateness.

The inverse of engaging on the two dimensions that matter. Reassuring type is uniform, balanced, symmetrical and unremarkable. It's credible and it's forgettable, and those are the same decision.

Prominent

High weight. Low naturalness, low flourish, low harmony.

Prominence is almost entirely a weight decision. If a client wants to be impossible to miss, the honest answer is that you're going heavy, and going heavy costs you naturalness, flourish and harmony simultaneously. That's three of the six dimensions traded for one impression.

The trade-offs, stated plainly

Lay the four recipes over each other and the conflicts are structural:

Elaborateness is required for engaging and disqualifying for both pleasing and reassuring. You cannot be interesting and calm using the same dimension settings.

Harmony is required for pleasing and reassuring, and must be low for engaging and prominent. Half the outcomes need agreement between letterforms and half need friction.

Naturalness is high for pleasing and engaging, low for prominent. Warmth and volume pull against each other.

Weight only appears in one recipe. Prominence. Which means every time you reach for a heavier weight to make something feel more substantial, you're buying prominence and paying for it with whatever else you had.

The same trade-off, found independently

If one 2004 paper isn't enough, the interesting-versus-trustworthy conflict turns up again in a completely separate study.

Hagtvedt (2011), in the Journal of Marketing, tested incomplete typeface logos. Letters with parts deliberately missing or blanked out. The result: incomplete logos negatively influenced perceived trustworthiness, because of lower clarity, and positively influenced perceived innovativeness, because of greater interestingness. Same axis, opposite ends, one design decision.

There's a further wrinkle worth knowing if you sell anything health-adjacent. Incomplete logos negatively influenced overall attitude toward the firm specifically for consumers with a prevention focus rather than a promotion focus. Prevention-focused consumers are the ones buying to avoid a bad outcome. That is a very large share of the skincare market.

So the clever, cropped, deconstructed wordmark is not neutral. It buys innovation, sells trust, and disproportionately alienates the anxious buyer.

The genuinely useful application

Here's the part that changes how you build a system rather than how you pick a font.

Because impressions come from dimension settings rather than from specific typefaces, multiple different typefaces can produce the same impression profile. Which means brand consistency isn't one typeface everywhere. It's one impression profile everywhere.

You can run a different face on packaging, on the site, in a clinical insert and in an investor deck, and stay coherent, provided each one sits in the same profile. The system holds because the dimensions hold, not because the letterforms match.

That's a considerably better argument than the one most of us have made at clients, which is some version of please stop changing the font on Instagram.

What this looks like in beauty

The category tends to want engaging and reassuring at the same time, which the model says you can't have from a single typeface. The workaround is to split the job.

Buy engagement in the display face. High elaborateness, low harmony, compressed. That's where distinctiveness is affordable, because it's operating at large sizes on a limited number of words.

Buy reassurance in the text face. High harmony, low elaborateness. Ingredient lists, directions, claims, anything a prevention-focused buyer reads carefully before deciding.

Buy prominence deliberately or not at all. Weight is the single lever, and it's expensive. Every step heavier costs naturalness and harmony, which means it costs warmth and calm.

The failure mode I see constantly is a brand that has bought engagement in every face, including the text one, and cannot understand why the packaging reads as untrustworthy despite testing well on "looks premium."

Caveats, since we're doing that now

It's 2004. Pre-variable fonts, pre-web-font explosion. The dimensions should be stable because they're perceptual rather than technological, but the stimulus set was of its time.

The sample and stimuli were corporate and US-based. Cross-cultural work since suggests the dimensions travel further than the impression mappings do. What reads as reassuring in Melbourne isn't guaranteed to read that way in Tokyo.

I have not read the full paper. I have the abstract, the factor loadings via a doctoral thesis that quotes them directly, and the four impression recipes via a summary table in a 2018 Journal of Consumer Research paper. Every specific claim above comes from one of those. The granular per-characteristic coefficients are on page 70 of the original, behind a paywall, and if you're going to quote a number rather than a direction, get the PDF.

The one-line version

Typography is a budget, not a wish list. Pleasing, engaging, reassuring and prominent are four different spends, and the currency is the same six dimensions.

Decide which one you're actually buying before you open a type foundry's website.