Most resumes that get filtered out are not filtered because the candidate was unqualified. They are filtered because of formatting or wording choices that confuse the software reading them before a person ever does. Understanding how that software actually works makes these mistakes easier to avoid.
How an ATS actually processes your resume
When you submit a resume, an applicant tracking system extracts your contact details, work history, education, and skills using text-parsing technology built to recognize patterns in resume structure. It then compares what it extracted against the job description and other criteria the employer set, looking for keyword and phrase matches. Some systems do simple keyword matching, checking whether a specific phrase appears at all. More advanced systems attempt semantic matching, recognizing that "PM" and "project manager" refer to the same role. You cannot know in advance which type of system a given company uses, which is exactly why the safest formatting choices are the ones that work regardless of sophistication.
Using tables, columns, or text boxes
An ATS reads resumes as a stream of text, not a visual layout. When you use a table or a multi-column design, the parsing order can become unpredictable, since the system may read across rows or columns in a sequence that scrambles your actual job history. Text boxes are a related risk, since some parsers skip content placed inside them entirely.
Saving section headers as anything other than the expected words
An ATS looks for standard headers like Experience, Education, and Skills to know where each type of information begins and ends. A creative header such as "Where I've Been" instead of "Experience" can cause the parser to miss that entire section, even though a human reader would understand it immediately.
Putting contact information in a header or footer
Some parsing systems do not read page headers and footers at all, treating them as decorative rather than content. If your email and phone number live only in a header, they may never make it into the system's extracted data, which means a recruiter searching the database by contact info will not find you, even if your resume technically exists in their system.
Listing duties instead of results
This one is not a parsing problem, it is a scoring problem. A phrase like "Responsible for managing a team" describes a role, not an achievement, and gives the ATS's keyword and relevance matching little to work with beyond the word "managing." A stronger version states what you actually did and what happened as a result, ideally with a number attached, since specific, results-oriented language tends to contain more of the actual terms a job description is looking for.
Keyword mismatch with the job description
If a posting asks for "project management" and your resume only ever says "managed projects," a simple keyword-matching system may not register that as a match, even though the meaning is functionally identical to a human. This is not about keyword stuffing, which creates its own problems. It is about matching the exact phrasing a specific posting uses, wherever that phrasing accurately reflects your real experience.
Using an image-based PDF or an unusual file type means an ATS without OCR capability cannot extract any text from your resume at all. It can look perfect on screen and still be functionally invisible to the system reading it. Check by trying to highlight or copy the text in your own exported file. If you cannot select individual words, the ATS likely cannot either.
Why these mistakes are easy to miss
Every one of these problems can look completely fine to you and to anyone you ask to review your resume by eye, since a human brain fills in structure and context automatically in a way that parsing software does not. The gap between how your resume looks and how it actually gets read by a machine is exactly where qualified candidates lose ground before anyone reviews their actual experience.