Resume Templates Guide for 2026: How to Choose an ATS-Compatible Format That Reads Well

The four structural choices that decide ATS-compatibility

Resume design contains many decisions; only four affect ATS parsing materially. The rest is taste.

1. Column count. Single-column is the safe choice across ATS deployments. Two-column templates fail unpredictably; Greenhouse's own resume parsing documentation lists columned layouts among the formatting choices that can produce partial or failed parsing, and the failure modes vary across older deployments and across non-major vendors. The design benefit (visual scan-ability) does not outweigh the parsing risk for tech-company applications. Reach for two-column only when applying to design roles where the resume is itself a portfolio artifact, and only after testing the export.
2. Font system. Common sans-serif at 10-11 point body, 12-14 point section headers. Helvetica, Arial, Calibri, and Source Sans Pro are widely supported across ATS deployments; font parsing still varies by export pipeline, embedding choice, PDF generator, and parser version, so the verification step (select-and-copy every text region in the export) is non-optional regardless of font choice. The trap fonts in 2026: Times New Roman (reads dated), Garamond (variable rendering), Inter and SF Pro (newer; inconsistent ATS support), Comic Sans and other decorative fonts (unprofessional). The font is rarely the decisive signal; the absence of a glaring font choice is.
3. Section order. Header (name, contact, links), professional summary or skills snapshot, experience (reverse-chronological), education, skills, optional (publications, talks, volunteer). Most modern ATS parsers handle reasonable section orders, but recruiters scan in this order; deviations cost you the 6-8 second window. The exception: new graduates or career-transitioners list education before experience because it is the more recent and relevant signal.
4. Export quality. The PDF must contain real text, not images of text. Every export pipeline can produce a flat-image PDF accidentally: Canva sometimes, Figma commonly, design-heavy Word templates rarely, and any manual scan-to-PDF always. Verify by opening in Preview or Acrobat and confirming you can select-and-copy every text region.

Format choices: Word, PDF, LaTeX, markdown

Most US tech-company applications today accept both PDF and Word uploads; PDF has become the more common default in practice, with Word retained for pipelines that need to edit the document downstream. Greenhouse and Lever both document support for PDF and DOCX uploads.¹²

• Microsoft Word (DOCX). Best for editability; some companies route the DOCX through internal pipelines that need to edit it. Trade-off: rendering fidelity varies across Word versions and OS, so the recruiter may see a slightly different layout than you authored. Conservative templates (Calibri body, single column, standard styles) render most reliably.
• PDF from Word export. Our recommended combination for most candidates. Author in Word, export to PDF, verify select-and-copy works on the export. Greenhouse and Lever both document support for PDF and DOCX; Lever explicitly recommends the highlight / select-and-copy test, and Greenhouse documents image uploads and inconsistent formatting as common parse-failure causes.¹²
• LaTeX. Excellent typographic output if you are comfortable with the toolchain. The LaTeX Project hosts the canonical distribution; moderncv and altacv are well-known LaTeX resume document classes.⁵ Trade-off: PDF only (no DOCX export), and the learning curve is real if you do not already know LaTeX.
• Markdown to multi-format via Pandoc. Pandoc exports markdown source to both PDF (via LaTeX) and DOCX from a single resume.md file.⁶ Power-user choice; produces clean output but requires a working LaTeX install for the PDF path.
• Canva, Figma, design-tool templates. Visually appealing but parsing-risky. The export-to-PDF pipeline often flattens text regions to images, which is unparseable. Reach for these only for design-track roles AND only after verifying the export.

Length and structure: general guidance

Harvard OCS and Stanford Career Education guidance both converge on one page for early-career and most mid-career candidates, with two pages reserved for advanced or extensive experience. The level-by-level breakout below is a working interpretation for tech-track applications, not a direct quote from those sources.

• Junior (0-3 years), Mid (3-5 years), Senior IC (5-8 years). One page. Working interpretation: the discipline of fitting senior IC experience into one page demonstrates curation; resumes that overflow at this level read as resume-by-accumulation.³⁴
• Staff IC (8-12 years), Principal IC (12-20+). One to two pages. The two-page allowance opens because levels-equivalent context (Meta E6, Google L6, Stripe L5+) takes structured space to convey accurately. Two pages is the ceiling, not a starting target.
• Engineering Manager / Senior Manager / Director. Two pages. Management content (team scope, hiring evidence, OKR ownership, cross-functional partnership) requires the additional space; one-page management resumes systematically under-convey scope.
• VP / SVP. Two pages or a separate two-page brief. The one-page-vs-two-page debate matters less at this level; what matters is the directness of impact and the named scope of teams.

ATS vendor reality in 2026

A handful of ATS systems show up most often in tech-company hiring. Greenhouse and Lever are widely used at FAANG-tier and SaaS-tier (verifiable from public job-application URLs); Workday is common at large enterprise. Ashby has grown at AI-labs and growth-stage startups (verified examples include OpenAI, Linear, and Ramp; some other AI-labs and growth-stage companies use Greenhouse or Lever instead). Phenom and Eightfold appear at enterprise scale (verified examples: Microsoft surfaces Eightfold, Cisco surfaces Phenom). Parsing behavior across these systems is individually documented or vendor-published; the items below summarize publicly available guidance and our own cross-vendor observations:

• Greenhouse. Mature parsing across PDF and DOCX. Support documentation lists image uploads, columned layouts, unclear sections, and inconsistent formatting across sections as common causes of partial or failed parsing; the conservative inference is that single-column layouts with clearly delimited sections and consistent formatting parse most reliably.¹
• Lever. Similar parsing maturity to Greenhouse. Help documentation covers candidate-side resume parsing quality.²
• Workday. Heavily customized per customer, so parsing behavior varies more than at Greenhouse or Lever in our cross-vendor observations. Conservative templates (single-column, standard sections) parse reliably across the deployments we have seen; designer-heavy templates produce more variable results.
• Ashby. Modern parser; in our observations it handles PDF and DOCX uploads from conservative templates without obvious issue. Two-column behavior is closer to Greenhouse / Lever than to older Workday deployments, but cross-vendor consistency is not guaranteed.
• Phenom and Eightfold. Enterprise systems with their own AI-driven parsing pipelines. From candidate-facing experience, standard formats process without obvious issue at the upload stage; we cannot verify parser-internal quality without vendor-published documentation.

The conservative cross-vendor choice in 2026 remains single-column PDF-from-Word with common sans-serif font. Templates that pass that bar parse cleanly more often than not across the major vendors; the verification step (select-and-copy every text region in the export) catches the residual failure cases before submission.

Visual traps that backfire

1. Text inside images. The single most common parser-killer. Logos, infographics, icon-with-label pairings where the label is rendered as part of the icon. Fix: every text region must be selectable as text in Preview or Acrobat.
2. Tables for layout. Word and Google Docs make it easy to build a two-column resume using a table. Some ATS parsers extract table cells out of order, producing scrambled text. If you must use a table, single-column only and verify the export.
3. Headers and footers in Word. The Word 'header' region (top of page) and 'footer' region (bottom) sometimes get extracted separately or skipped entirely depending on the parser. Put your name and contact in the body, not in the header region.
4. Custom bullet glyphs. Standard round and square bullets parse cleanly. Decorative bullets (arrows, stars, custom glyphs) sometimes render as boxes or get dropped, breaking sentence flow on extraction.
5. Multi-color text and gradient fills. Visual interest, but parsers normalize to plain text and the gradient is wasted. Worse, low-contrast accent colors produce low-confidence text recognition in some parsing pipelines.
6. Photos in US / UK / Canada / Australia applications. Reject some ATS configurations on bias-protection grounds; in others they parse but the recruiter rejects on hiring-policy grounds. Continental European norms differ; know your audience.

From template choice to shipped resume

Template selection is the structural step; the content is the editorial step. The role-specific resume guides linked below cover the content layer for each major track. The sequence:

1. Choose a single-column template (Word or LaTeX).
2. Set the type system: common sans-serif, 10-11 point body, 12-14 point section headers.
3. Author content in role-specific terms (the role hub for your track is the canonical reference; cross-link below).
4. Export to PDF.
5. Verify select-and-copy works for every text region.
6. Run a parser-simulator if available; submit to a real application and request feedback if your ATS exposes it.

The verification step (point 5) is the single highest-impact 30 seconds in the resume process; skipping it is the most common single failure mode.