Original test
We tested 16 resume templates in 3 parsers. Nothing broke.
We ran every ResumeFinch template through three text extraction engines, with three synthetic resumes and two file formats. That is 144 runs and 5,280 checks.
Every check passed. No name, email, role, date, bullet or skill went missing, and every section stayed in order. The only measurable difference was contact placement in two sidebar templates.
144
parser runs
5,280
checks passed
100%
average score
100%
sections in order
Tested . No real resumes or personal data were used.
Why we ran the test
The ATS template page makes a format claim: one column is safer, and a second column can be read out of order by older systems. That claim is common advice, but it is rarely checked against the files the site actually exports.
So we checked. We generated the same PDF and Word files a customer downloads, then read them with parsers that do not share code. The question was narrow: does the text survive, and does it come out in a readable order?
This is a text extraction test, not a test of a named applicant tracking system. It shows what a parser can see. It does not show how every vendor maps those words into fields.
Method
We used all 16 ResumeFinch templates, three synthetic resumes and two file formats. Each format went through one or two extraction paths, for 144 runs in total.
| Input | What we used |
|---|---|
| Templates | All 16, from one-column Harbor to the Lumen and Anchor sidebars |
| Resumes | Early career, experienced, and career change, all fictional |
| Formats | PDF and DOCX, generated by the same export code the site uses |
| PDF engine | pdfjs-dist 6.4.299 |
| Word engines | Mammoth 1.13.0 and the DOCX XML text reader used by the export tests |
| Checks | Name, contact details, headline, summary, headings, titles, employers, dates, bullets, skills and line order |
Each check is pass or fail. The score is the share of checks that passed, so 100 means every field and every ordered line was found. The raw CSV has one row per run.
Results
All 144 runs scored 100. All 100% kept every section in order. The only variation was where the parser placed contact details in two templates.
| Format | Runs | Average score | Sections in order | Contact after experience |
|---|---|---|---|---|
| 48 | 100% | 100% | 12.5% | |
| Word | 96 | 100% | 100% | 0% |
| Template class | Runs | Average score | Sections in order | Contact after experience |
|---|---|---|---|---|
| ATS-safe | 90 | 100% | 100% | 0% |
| Design-forward | 54 | 100% | 100% | 11.1% |
| Template | Layout | Class | Runs | Score | Contact after experience |
|---|---|---|---|---|---|
| Harbor | One column | ATS-safe | 9 | 100% | 0 |
| Meridian | One column | ATS-safe | 9 | 100% | 0 |
| Slate | One column | ATS-safe | 9 | 100% | 0 |
| Summit | One column | ATS-safe | 9 | 100% | 0 |
| Keystone | One column | ATS-safe | 9 | 100% | 0 |
| Campus | One column | ATS-safe | 9 | 100% | 0 |
| Beacon | One column | ATS-safe | 9 | 100% | 0 |
| Regent | One column | ATS-safe | 9 | 100% | 0 |
| Linen | One column | ATS-safe | 9 | 100% | 0 |
| Ledger | Dates in margin | Design-forward | 9 | 100% | 0 |
| Atlas | Left sidebar | Design-forward | 9 | 100% | 0 |
| Lumen | Left sidebar | Design-forward | 9 | 100% | 3 of 9 |
| Vista | Right sidebar | Design-forward | 9 | 100% | 0 |
| Anchor | Left sidebar | Design-forward | 9 | 100% | 3 of 9 |
| Crest | Right sidebar | Design-forward | 9 | 100% | 0 |
| Banner | One column | ATS-safe | 9 | 100% | 0 |
What broke
Nothing broke. That is the finding, and it is worth stating plainly because the usual advice says the opposite.
In all 144 runs, the parser found every field. In all 144 runs, the sections appeared in the intended order. Word was clean in both extraction engines. PDF was clean in the PDF engine.
The one difference was contact placement. Lumen and Anchor keep contact details in a sidebar. The PDF renderer draws the main column first and the sidebar second, so the parser read the email and phone after the experience section. The same files in Word move those details under the name, where they read first.
That is an order difference, not a text loss. It matters because an older system may expect contact details near the top. It does not mean the resume was unreadable.
What this means for job seekers
- Put your email and phone in the header, not only in a sidebar, when you apply through a large employer portal.
- Choose the Word download when you cannot test the exact file. It is one column by design.
- Do not assume a two-column PDF is broken. In these parsers it was not, but the sidebar was read after the main column.
- Use standard headings such as Summary, Experience, Education and Skills. Every parser found them.
- Test your own file with the free checker before you apply. It reads the file in your browser and shows what a simple parser can see.
Limitations
This test used three open-source extraction engines, not a named applicant tracking system. It did not test field mapping, ranking, or any vendor's parsing rules.
The resumes are synthetic and the files are US Letter. Different fonts, longer resumes, graphics, tables, text boxes and page breaks can change what a parser sees.
The results describe these 16 templates and these three resumes on the test date. Parsers change, and a future version may read the same file differently.
A clean extraction does not guarantee a job interview. It only means the text was there to read.
Frequently asked questions
Do resume templates pass ATS?
In this test, yes. All 16 ResumeFinch templates kept every field in all 144 parser runs. That does not guarantee that every applicant tracking system will map every field correctly, because vendors differ and field mapping is not the same as text extraction.
Is PDF or Word better for an ATS?
Word is the lower-risk default. Every Word download is one column and passed every check. PDF passed every check too, but Lumen and Anchor put contact details in a sidebar, so the PDF parser read those details after the experience section.
Are two-column resumes readable?
In these three parsers, yes. The sidebar was read after the main column, not woven through it. Older systems can still read columns differently, so a one-column template remains the safer choice for large employer portals.
Did any template lose text?
No. No name, email, phone, headline, summary, heading, job title, employer, date, bullet or skill went missing in any of the 144 runs.
Which template had the best result?
All 16 templates scored 100 on the 5,280 checks. Lumen and Anchor were the only templates where contact details came after experience in PDF extraction. That is an order difference, not a missing field.
Can I test my own resume?
Yes. The free resume checker reads your file in your browser and shows what a simple parser can see. It does not upload or store the file.
Cite this study
ResumeFinch. (2026, November 5). We tested 16 resume templates in 3 parsers. https://resumefinch.com/research/ats-resume-templates
The full dataset is available as a CSV download. The test code lives in the ResumeFinch repository under research/ats.
Start with a template that passed
Build a resume free, preview every change, then download a PDF and a Word file.