Most Resume Formatting Advice Fails a Basic ATS Parsing Test. Here Is What Survives.

Before an ATS scoring engine evaluates your keywords, skills, or experience, something else happens first: parsing. The system has to read your resume file, extract the text, identify section boundaries, and structure the content into fields it can score against. If the parsing step fails or produces garbled output, everything downstream suffers. Keywords get lost. Section headers are misidentified. Experience entries bleed into each other.

Most resume formatting advice focuses on what looks good to a human reader. But the formatting choices that make a resume visually appealing are often the same ones that create parsing failures. We examined this problem using data from our production pipeline, where we see the actual parsed output that ATS engines work with, not just the final score. The gap between what candidates submit and what the system actually reads is larger than most people realize.

Parsing and scoring are separate steps, and the first one matters more than you think

ATS scoring is not a single operation. It is a pipeline with at least two stages. Stage one is parsing: extracting text from the document, identifying structural elements (section headers, bullet points, dates, job titles), and organizing the content into a structured representation. Stage two is scoring: evaluating the parsed content against the job description across the five components (keywords, skills, experience, education, contextual fit).

Most candidates focus entirely on stage two. They optimize keywords, rephrase bullet points, and tailor their content to each job. But if stage one produces a degraded version of their resume, the stage two optimizations are working with corrupted input. You can have perfect keyword coverage in your actual document, but if the parser cannot extract those keywords because they are trapped in a text box or rendered as an image, the scoring engine will never see them.

The formatting choices that cause the most parsing damage

Not all formatting problems are equal. Some cause minor information loss. Others render the resume nearly unreadable to the parser. Based on what we have observed across our production pipeline and targeted format testing, we can rank the most common formatting choices by their impact on parsing quality.

The pattern is clear: the more visually creative the formatting, the higher the parsing risk. This creates an uncomfortable tension for candidates. The templates that stand out on a computer screen are often the ones that perform worst through ATS parsing. And because parsing failures are invisible to the candidate (you never see the parsed output), the damage goes undetected until the rejection arrives.

How parsing damage cascades into scoring

Parsing problems do not just cause generic "lower scores." They damage specific scoring components in predictable ways. Understanding which components are most vulnerable helps explain why two candidates with similar qualifications can receive dramatically different scores based solely on their resume formatting.

The vulnerability pattern mirrors the weight distribution. Keywords, carrying the highest weight at 40%, are also the most vulnerable to parsing failures. This is particularly damaging because, as we showed in our before-and-after analysis, keyword improvement is the primary mechanism through which optimization works. If parsing has already corrupted the keyword layer, optimization has less material to work with.

What a parsing-safe resume actually looks like

The resumes in our production dataset that scored best before optimization shared structural characteristics that had nothing to do with their content quality. They were formatted in ways that minimized parsing ambiguity. Here is what they had in common.

None of these characteristics make a resume look better to a human recruiter. Some of them actively make it look plainer. But they ensure that the parser delivers a clean, accurate representation of the resume's content to the scoring engine. In ATS processing, boring formatting is reliable formatting.

The score impact of format choices in our data

Among the 22 base resumes in our dataset, the ones with clean single-column formatting consistently outperformed those with more complex layouts on the keyword component, even when their actual keyword coverage was comparable. This is not a content effect. It is a parsing effect. The well-formatted resumes simply delivered more of their content to the scorer intact.

We also tested reformatting several resumes from their original complex layouts into clean single-column versions, keeping the content identical. The results were consistent: the reformatted versions scored higher on keywords, and in some cases the improvement was substantial enough to move the overall score up by several points before any content optimization was applied.

The two-column sidebar case is instructive. The candidate had listed relevant technical skills in a sidebar column. When the parser processed the resume, it merged the sidebar text with the main column, producing jumbled output where "Python" appeared mid-sentence in an unrelated experience bullet. The keyword scorer could not match "Python" because it was no longer in a recognizable context. Simply moving the skills to a dedicated section in the main column, with no other changes, boosted the keyword score from 14 to 29.

The file format question: PDF versus DOCX

This is one of the most frequently asked questions in resume optimization, and the answer is more nuanced than the usual advice suggests. Both PDF and DOCX are acceptable for most ATS systems in 2026. But they have different parsing characteristics that matter in edge cases.

In practice, the difference between a well-made PDF and a well-made DOCX is minimal for parsing purposes. The problems arise from how the document was created, not which format it was saved in. A PDF exported from Google Docs with single-column formatting will parse cleanly. A PDF exported from Canva with multi-column layouts, custom fonts, and graphic elements will not. The file extension is less important than the creation method.

The format decision framework

Given the data, the formatting decision comes down to a simple question: are you optimizing for human eyes first or ATS parsing first? In most hiring pipelines today, the ATS sees your resume before any human does. That suggests a clear priority order.

Step five is the most practical advice in this article, and it is something almost no one does. Open your resume in Preview or Adobe Reader, select all text, and paste it into Notepad or a plain text editor. What you see is roughly what the parser will see. If your skills section is jumbled, if dates appear next to the wrong job titles, or if entire sections are missing, you have a formatting problem that will affect your ATS score before you write a single word.

How this connects to our broader scoring research

In our previous articles, we have documented the scoring patterns that emerge from scoring the same resume against different jobs, the mechanics of optimization, and the differences across career levels. All of that research assumes clean parsing. The scores we reported came from resumes that had been properly parsed.

For candidates whose resumes have formatting issues, the actual scores will be lower than what our research would predict for their content quality. The score gap between a well-formatted and poorly-formatted version of the same resume is not a content problem or a keyword problem. It is a parsing problem, and it happens before the scoring engine even runs.

Fixing formatting is therefore the highest-leverage, lowest-effort change most candidates can make. It costs nothing, requires no content changes, and removes a silent penalty that may be suppressing their scores by several points. For candidates who have already optimized their content and are wondering why their scores are not improving, the format is the first thing to check.