Six free resume templates, one resume, 30 rules

The results

Every row is a real PDF you can open above, and re-score yourself on the free checker. A rule marked fail costs its full weight; a warning costs part of it.

• •Elegant. A photo slot and a two-column skills rail; the summary and the skills block do not parse as sections.
• •Flat. Runs the same content to three pages; dates and bullets are laid out in a form the parser does not read as such.
• •Stack Overflow. Icon glyphs in the contact line and letter-spaced headings; bullets lose their verbs to the icon prefix.
• •Class. Dates sit in a side column and detach from their roles when the text is extracted.
• •Kendall. A banner header the parser reads after the body; the education block is not found at all.
• •Paper. A photo and a two-column header; the skills section is not recognised and the contact line lands mid-page.

How we did it

• •One resume. The customer-success sample that ships with CVLoom (three roles, measured bullets, grouped skills, two certifications), converted to the JSON Resume format the themes read. The file is published beside the PDFs as resume.json.
• •Their renderer, not ours. Each theme rendered the resume with its own code, unmodified, at its own defaults. We printed the result to PDF with the same headless Chromium that exports every CVLoom document, US Letter, half-inch margins.
• •The same 30 checks, from the PDF. Text extracted the way a parser extracts it, then the published rules. No hand scoring. The script is in the CVLoom repository as scripts/research-free-templates.mts.
• •Why these six. They are free, downloadable without an account, and rendered by open code, so the test is reproducible. A seventh theme, Even, would not render outside its own toolchain and was left out. Microsoft Word's templates download freely too, but printing them faithfully needs Word itself, and we could not automate that cleanly; they are the next batch.

Why our 96 is not a fair fight

CVLoom's templates were designed against these rules and re-measured every time a rule changed, so scoring well on them proves consistency, not superiority. The honest reading of the table is narrower: the same content loses 5 to 13 points to layout alone. Two-column headers put the phone number where a parser reads it last; date columns detach dates from roles; photo slots and icon glyphs add elements no field can hold; a three-page render fails the length rule outright. Those are the things a template decides for you, and the reason the rules are public is so you can check any template, including ours, for them.

What we could not score

The original plan was to score the paid builders' own sample PDFs. On 1 September 2026 we opened each builder's templates and examples galleries looking for one. None publishes any: every one shows images of resumes, not resumes. That finding stands and is the reason this study uses free templates instead.