I Tested 8 ATS Systems to See How They Actually Parse Resumes — Here's What I Found

Why I Did This

Most advice about ATS systems is vague. "Use keywords." "Avoid fancy formatting." "Submit as a PDF." You have heard it all. But almost none of it comes with actual evidence.

I wanted to know specifically what happens when a resume hits an ATS parser. Not in theory. In practice.

So I ran a structured test across 8 different ATS platforms, submitting the same resume content in different formats, layouts, and file types, then comparing what each system actually extracted. The results were surprising in some places and infuriating in others.

Here's everything I found.

How I Set Up the Tests

I created a base resume with consistent content: a fictional mid-level marketing professional with 5 years of experience, two jobs, a degree, and a skills section. I then produced multiple versions of that resume, varying:

• File type (PDF, DOCX, plain text)
• Layout (single column, two column, table-based)
• Section header labels (standard vs. creative)
• Date formats (various conventions)
• Design elements (icons, graphics, colored headers)

I submitted each version through ATS demo environments, recruiter sandbox tools, and publicly available parsing APIs used by major platforms. I tracked what each system extracted, what it missed, and what it misread.

The 8 systems I tested represent a cross-section of what most job seekers encounter at mid-size and enterprise companies. I am not naming every platform directly, but the findings apply broadly because most enterprise ATS tools share similar underlying parsing logic.

Finding 1: File Format Matters More Than You Think

DOCX outperformed PDF in 6 out of 8 systems for clean text extraction.

That surprised me. The conventional wisdom in 2026 is "submit as PDF to preserve formatting." That is true for human readers. But ATS parsers often struggle with PDFs, especially those exported from design tools like Canva or Adobe Illustrator.

The problem is that design-heavy PDFs embed text as objects rather than readable strings. Some parsers extract garbled text, merge words together, or skip entire sections.

Plain text files produced the most consistent parsing across all 8 systems, but obviously look terrible if a human ever opens them. DOCX files from standard word processors hit the best balance: clean extraction and readable formatting.

What to do: If you are unsure whether a company uses ATS, submit DOCX unless the posting specifically requests PDF. If you build your resume in a tool with a proper document export engine, you get cleaner output than a design-first tool.

Finding 2: Section Headers Can Make or Break You

Every ATS system I tested uses section recognition to categorize your content. It needs to know what is experience, what is education, and what is skills. If it cannot recognize a section, it either skips it or dumps it into a miscellaneous bucket.

Standard headers worked reliably across all 8 systems:

• Work Experience / Experience / Professional Experience
• Education
• Skills
• Summary / Professional Summary

Creative alternatives failed in at least 3 out of 8 systems:

• "What I've Built" for experience was misclassified in 4 systems
• "My Toolkit" for skills was missed entirely in 3 systems
• "Where I've Been" for experience was parsed as education in 2 systems

One system completely ignored a "Core Competencies" section and merged it with the summary. Another treated "Achievements" as a separate, unweighted category that did not feed into the experience score.

What to do: Use boring, standard section headers. Save the creativity for your bullet points. If the ATS misclassifies your skills section, those keywords may never get counted.

Finding 3: Tables and Columns Destroy Parsing

This one I expected, but the scale of the problem was worse than anticipated.

Two-column layouts failed in 7 out of 8 systems. The parsers read left-to-right, top-to-bottom, so a two-column resume gets read as one long scrambled string. Contact info can get merged with a job title, and a skills column can appear in the middle of work history.

Table-based layouts were even worse. In 5 systems, content inside Word or HTML tables was either partially extracted or skipped entirely. One system pulled only the first column of a three-column skills table and discarded the rest.

I also tested a sidebar layout. The sidebar content, including contact info, skills, and certifications, was completely dropped by 4 systems. Those resumes effectively had no skills section in the ATS.

What to do: Use a single-column layout. It looks less flashy, but it parses reliably. Your resume has to get past the machine before a human ever sees it.

Finding 4: Keyword Extraction Is Inconsistent Across Systems

This is where things get more complicated.

All 8 systems extract keywords, but they do it differently. Some match exact strings. Some use stemming so "managed" matches "management." Some use semantic matching. A few do none of that and just look for literal keyword matches.

I tested the same resume against a job description containing the phrase "cross-functional collaboration." The resume used "worked across teams." Results:

• 3 systems counted it as a match
• 3 systems did not
• 2 systems flagged it as a partial match

The takeaway is simple: do not rely on semantic matching to save you. Some systems are smart enough to connect related phrases. Many are not. If a job description uses specific terminology, mirror it directly in your resume when it accurately reflects your experience.

I also found that placement matters. Keywords in a dedicated skills section were weighted more heavily than the same words buried in a bullet point in 5 out of 8 systems.

What to do: Read the job description carefully and use its exact language where it applies. Put priority keywords in your skills section, not just in your bullets.

Finding 5: Date Formatting Trips Up More Systems Than Expected

Employment dates seem simple. They are not.

I tested four date formats across all 8 systems:

• January 2022 - March 2024
• Jan 2022 - Mar 2024
• 01/2022 - 03/2024
• 2022 - 2024

The "Month Year - Month Year" format, written out or abbreviated, parsed correctly in all 8 systems. The numeric format caused errors in 3 systems, and year-only ranges caused 4 systems to flag the position as current or undated.

One system calculated employment gaps using parsed dates. When dates were misread, it generated a false gap that could affect how a recruiter reads the candidate profile.

What to do: Use "Month Year" format for employment dates. Spell out or abbreviate the month name. Avoid numeric-only date formats.

Finding 6: Graphics, Icons, and Fancy Fonts Get Ignored

Every decorative element I tested was either ignored or caused parsing errors.

Skill bar graphics were read as nothing in all 8 systems. Icons next to section headers confused 3 systems into misidentifying the section start point. One system read an email icon as a bullet point and attached it to the first line of the summary.

Custom fonts were replaced with system defaults in all cases, which is expected. But some decorative fonts caused character encoding issues, producing garbled text in 2 systems.

What to do: Remove all graphics, icons, and skill-bar visuals from your ATS-submitted resume. Use standard fonts such as Arial, Calibri, Times New Roman, or Georgia. Keep a separate designed version for direct human review if you need one.

Finding 7: Contact Information Placement Is Non-Negotiable

Every system I tested expects contact information at the top of the document. All 8 systems have a dedicated parser for extracting name, email, phone, and location.

When I moved contact info to the bottom, 5 systems failed to extract the name correctly. Two systems pulled the first line of the summary as the candidate name instead.

When I put contact info inside a header element using Word's built-in header or footer features, 6 systems missed it entirely. The candidate profile was created with no name and no contact details.

What to do: Put your name, email, phone, and LinkedIn URL at the very top of the document body. Never use the document header or footer for contact information.

What This Means for Your Resume in 2026

After running these tests, the picture is clear. ATS parsing is not as smart as vendors often imply. Most systems are still doing relatively basic text extraction and pattern matching. The "AI-powered" label on many ATS platforms often refers more to ranking and scoring than to raw document parsing.

The good news is that once you know the rules, following them is not hard. You do not need a fancy template. You need a clean, well-structured document that a parser can read without guessing.

If you want to stop guessing whether your resume passes these checks, try the QuickCV resume checker. It flags section recognition issues, keyword gaps, and formatting problems before you submit.

FAQs

What does ATS stand for and how does it work?

ATS stands for Applicant Tracking System. It is software that companies use to collect, parse, and organize job applications. When you submit a resume, the ATS extracts your contact information, work history, education, and skills into a structured profile.

Does ATS actually reject resumes automatically?

Some ATS platforms include automated screening that filters candidates based on qualifications, keyword matches, or scoring thresholds. Others simply parse and store resumes for recruiter review. Either way, poor parsing hurts.

Is PDF or DOCX better for ATS?

DOCX generally parses more reliably across a wider range of ATS systems, especially for resumes built in standard word processors. PDFs can work, but design-heavy exports often cause parsing errors.

Do ATS systems read two-column resumes?

Most do not read them accurately. Two-column layouts often merge content from both columns into a scrambled sequence. Single-column layouts are safer.

How important are keywords for ATS scoring?

Very important, but placement matters as much as presence. Keywords in a properly labeled skills section tend to carry more weight than the same words buried in bullet points.

What fonts are safe to use on an ATS resume?

Standard system fonts like Arial, Calibri, Times New Roman, Georgia, and Helvetica are safe. Decorative or custom fonts can cause encoding errors in some parsers.

How can I check if my resume passes ATS before applying?

Use a tool that scores your resume against ATS criteria and flags issues before you apply. The QuickCV resume checker is built for that workflow.