1. Why we ran this test
Almost every resume builder calls its templates "ATS-friendly", and almost none show evidence. An applicant tracking system starts by turning your PDF into plain text. If that step loses your phone number, scrambles your jobs or merges a bullet into your skills, no amount of keyword work fixes it. We wanted to know, template by template, what survives that first step, and to publish the answer whether it flattered us or not.
2. Method
- One standard resume. A fictional candidate, Jordan Ellis, with a name, job title, email, phone, city, LinkedIn URL, a short summary, two jobs with five bullets in total, one degree and eight skills. The same data went into every template.
- The real export path. Each template was opened in the live builder in Google Chrome 153 and saved as an A4 PDF through the same print path a user gets from the Download PDF button.
- Two extractors. Every PDF was read with Poppler's pdftotext 24.02, the engine behind many server-side tools, and with pdfminer.six, a Python library used inside several open-source resume parsers. Both ran with default settings.
- Automatic checks. On each extracted text we checked whether there is a text layer; whether the name is in the first lines; whether the email, phone and location are present and unbroken; whether the section headings are recognisable (Experience, Education, Skills or their usual variants); whether the two jobs appear in the right order; whether each of the five bullets is intact; whether each job's dates appear within 250 characters of its employer; whether all eight skills are present; and whether any characters are unreadable.
- Scope. 72 resume designs and 8 cover letter designs, tested on 14 November 2026. Cover letters were checked only for a text layer, the name, the email and the letter text.
3. Results at a glance
| Check (72 resume designs) | Poppler | pdfminer.six |
|---|---|---|
| Real text layer, one A4 page | 72 | 72 |
| Name at the top | 72 | 72 |
| Email, phone and location intact | 72 | 72 |
| Standard section headings found | 72 | 72 |
| Jobs in the correct order | 71 | 72 |
| All five bullets intact | 72 | 71 |
| All eight skills found | 72 | 72 |
| No unreadable characters | 72 | 72 |
| Each date next to its job | 64 | 53 |
43 templates passed every check in both extractors, dates included. 70 passed everything except the date check. All 8 cover letters exported with a text layer, the name, the email and the full letter text.
4. What we fixed because of the test
The first run was worse. The problems it found are worth knowing about, because they can affect resumes made with any tool:
- Unreadable dashes from web fonts. Chrome embedded our variable web fonts as "Type 3" fonts in the PDF. pdfminer.six could not map some of their characters, so the dash in date ranges such as "2021 – 2024" came out as garbage in 61 of 80 PDFs. We now ship static font files, and the count is zero.
- Small capitals split words. One academic template used small capitals for headings, and the extractors read "Profile" as "P rofile". Headings now use real capital letters.
- Wide letter spacing. A fashion-style template spaced the letters of the name so far apart that both tools read "J O R D A N". We reduced the spacing, and the name now extracts as one word.
- Non-standard headings. Two layouts used labels such as "Appointments" and "Research interests" on their own. They now say "Appointments and experience" and "Research interests and skills", which a parser can map.
5. The date problem
Many designs put the dates at the right end of the job title line. On screen the dates clearly belong to the job. In the PDF's reading order, though, an extractor may finish the left-hand text first and pick up the right-aligned dates later, sometimes after the bullets. A parser that links dates to jobs by position can then assign the wrong dates, or none, which can affect "years of experience" calculations.
This is a design trade-off rather than a bug, so we have not changed the designs silently. If dates matter for the roles you apply to, choose a template marked "Yes" in both date columns below, or check your own PDF with the copy-and-paste test in our guide to how ATS software reads resumes.
6. Results for every template
"Pass" means every check except the date check passed in that extractor. "Apart" means at least one date range was more than 250 characters from its employer in the extracted text. Click a template name to open it in the builder.
7. Limits of this test
- These are not commercial ATS products. Workday, Taleo, iCIMS, Greenhouse and others use their own parsers, which we cannot run. Text extraction is the first step they all share, so a template that fails here is at risk, but passing here does not guarantee how a particular employer's system will treat your resume.
- One resume, one language. The test used a short English resume. Long content, other languages and right-to-left scripts can behave differently.
- Automatic checks. The checks look for exact text. A person reading the output might accept small differences that a check counts as a miss, and the reverse.
- A snapshot. Browsers and templates change. We will rerun the test when we change the templates or the export path, and update this page.
Frequently asked questions
Which TrulyFreeCV templates are safest for ATS?
The 43 templates marked "Pass" in both extractor columns and "Yes" in both date columns passed every check. For the plainest reading order, single-column layouts such as Harvard Academic Classic ATS and Executive Corporate are a safe start.
Does a two-column template fail ATS?
Not in this test. Several two-column and sidebar designs, such as Dual-Column Modern Tech Sidebar, passed every check. Problems came from specific details, like right-aligned dates or decorative letter spacing, rather than from columns as such.
Can I run the same check on my own resume?
Yes. Open your PDF, select all, then copy and paste into a plain text editor. Your name should come first, each job should keep its title, employer and dates together, and every bullet should read as one sentence.
Why not test real ATS software?
Commercial systems do not offer public test access, and paid "ATS score" tools do not explain how they score. We used open-source extractors so that anyone can repeat the test with the same tools.
Sources and tools
- Poppler, PDF rendering library and the pdftotext utility.
- pdfminer.six, Python PDF text extraction library.