Résumé design: what matters, what doesn't

The résumé chooser in jobhunt.coach offers Classic, Accent and Modern. In the default HTML renderer, Classic centers a serif name, Accent uses warm brown headings, and Modern puts section labels in a narrow column beside the content. The differences are visible. I still wouldn't treat choosing between them as the part of an application that deserves the most time.

Building those layouts has meant fixing smaller, less glamorous things: employment dates wrapping onto two lines, headings sitting too close to the preceding section, and a short spill onto a second page.

The older full-authoring LaTeX path uses a written brief for each style, and the model writes the whole document against it. Those briefs remain in use for general-purpose LaTeX résumés. They show how much formatting instruction a model needs when it owns the entire page.

The retained Modern brief asks for a sidebar about a third of the page width and explicitly rejects equal columns. The Classic and Accent briefs specify two centered contact lines. The shared rules name the allowed icon package and say that a GitHub icon belongs only beside a GitHub link. These instructions govern the full-document LaTeX path.

Today, the default HTML path gives the model the body to write while the application supplies contact details and formatting. LaTeX remains an option; its job-specific résumés use fixed layouts by default too.

The useful design work is in making the structure predictable. A restrained page still needs clear headings and enough space between unrelated pieces of work. A plain page deserves that check too.

We tested the conservative direction deliberately, too. In July I prototyped a Word renderer and stripped it to the studs: no tables, no text boxes, no multi-column tricks, Word-native bullets instead of typed bullet characters, explicit fonts and spacing rather than whatever the local Word install happens to inherit. I described that structure as ATS-friendly in the findings doc, and I want to be careful about what that meant: friendly-shaped, not verified. We never measured it against vendor parsers, so I won't claim it clears anything. (The scanners post covers what that layer actually does.)

That prototype gave us a useful comparison. We rendered two samples through the same conservative layout and applied the same page-fill measurement. The sparse sample came in at 51%. A deeper fictional sample, three roles' worth of specific detail, reached 74%. Same type size. Same spacing. The difference in that comparison was the content.

Spacing matters too. Our current HTML renderer can adjust spacing when a first, single or trailing page has a large unused area. The same text can fit better without reducing the type size. In one saved comparison, compacting the layout kept a résumé on one page but left headings too close to the section above. Moving space from below the headings to above them made the grouping clearer, with the text and type sizes unchanged.

If the experience itself is thinly described, though, stretching the spacing won't supply the missing detail. Go back to what you actually did: what you owned, what you made, and what happened. That record deserves time before another pass through the font menu.

When checking a finished résumé, find the most recent role first. Check that dates sit consistently beside their entries and headings stay with the content they introduce. Our three styles differ in layout as well as typography and color, but each needs to pass those ordinary reading checks.

Open the exported PDF before sending it. Read the section transitions, follow the dates, and try the links. Once the page is easy to read, spend the next edit on what it says.