What an ATS actually does with your resume
An applicant tracking system stores applications for a company. When you apply, it extracts the text from your file and tries to fill fields like name, email, education and experience. Recruiters then search and filter that text, for example “React AND internship”, or skim a plain-text view. So two things matter:
- Can the text be extracted in the right order? If not, your details end up in the wrong fields or become gibberish.
- Does the text contain the words a recruiter will search for? If the job says “PostgreSQL” and your resume only says “databases”, a search won’t match.
Our test: one column vs two columns
On 3 October 2026 we took two resumes for fictional students, one in a single column and one with a skills sidebar, and ran them through two widely used open-source PDF text readers, pypdf and pdfplumber (the kind of library parsing software is built on). We checked whether the section headings came out in the right order and whether lines from different parts of the page got mixed together.
| Resume | pypdf | pdfplumber |
|---|---|---|
| One column (Res Do’s Jake-style PDF) | 4 of 4 headings found, in order; email found | 4 of 4 headings found, in order; email found |
| Same resume as Word (.docx) | 4 of 4 headings and the email read correctly (python-docx) | |
| Two columns (skills sidebar + main column) | Columns read separately (0 mixed lines) | 6 of 9 lines mixed sidebar and main text, e.g. “Python Data Intern, Acme Corp (2024)” |
The takeaway: the one-column resume came out correctly in every reader. The two-column resume depended on which reader was used: one handled it, the other merged a skill into a job title. You can’t know which kind of parser a company’s ATS uses, so one column is the safe choice. This is a small test with two readers, not a test of any specific commercial ATS.
What breaks resume parsing
- Scanned PDFs and photos of a resume: there’s no text to extract at all.
- Multiple columns and sidebars: as our test shows, some readers interleave them line by line.
- Text boxes, tables used for layout, and headers/footers: contact details placed in a Word header are skipped by some systems.
- Icons and skill bars: a phone icon instead of the word “Phone” is fine for people but meaningless to software, and a five-star “Python” rating carries no information.
- Unusual headings like “My Journey” instead of “Experience”.
Keywords: how to use them honestly
Read the job post and note the skills and tools it repeats. If you have them, use the exact wording in your skills line and, better, inside a bullet that shows you using it: “Built a REST API in Node.js and PostgreSQL…”. Don’t list skills you don’t have: you’ll be asked about them in the interview.
Res Do does this when you paste a job post: it works the job’s terms into your resume only where your resume or answers support them, and lists the rest separately as gaps.
ATS-friendly resume checklist
- One column, top to bottom.
- Standard headings: Education, Experience, Projects, Skills.
- Contact details in the main body, not in the page header.
- Real text: exported PDF or Word, never a scan or image.
- The job’s key terms, spelled the way the job post spells them, used where they’re true.
- Consistent dates (“Jun 2025 – Aug 2025”).
- No tables, text boxes, icons, photos or skill bars.
Every resume Res Do produces follows this list and is checked by extracting its text before you download it. If you want the format itself, see Jake’s Resume template or the fresher resume format.