How to Fix Your Resume for ATS (Without the Myths)

Almost everything written about applicant tracking systems is either exaggerated or simply wrong. You have probably read that 75% of resumes are "rejected by robots before a human sees them," that you should stuff keywords in white text, or that a particular font will get you filtered out.

None of that is accurate, and acting on it produces worse resumes. What is true is more mundane and more useful: ATS software is imperfect at reading documents, recruiters search those imperfect records using specific terms, and a resume that parses cleanly and uses the right vocabulary shows up in more searches than one that doesn't.

This guide covers how the software actually behaves, which formatting choices genuinely break it, how keyword matching really works, and how to test your own document before you send it.

What an ATS Actually Does

When you submit an application, the system does roughly four things in sequence:

1. Ingests the file and extracts raw text from it. This is the parsing stage, and it is where most damage happens.
2. Maps that text into fields — name, email, phone, employer, job title, start and end dates, education, skills. It uses section headings and layout position to guess which text belongs where.
3. Stores the record in a searchable database alongside every other applicant, and often alongside every applicant who has ever applied to that company.
4. Surfaces records to recruiters who search, filter, and sort. Some systems compute a match score against the posting; many do not, and where scores exist they usually rank rather than reject.

The critical point is step four. In the overwhelming majority of configurations, the software does not delete or auto-reject your application. A human recruiter searches "Salesforce administrator Boston" and works through the results. If your resume parsed badly, or uses different words than the ones they searched, you are in the database but you are not in the result set. Functionally identical to rejection, arrived at by a completely different mechanism.

You are not being rejected by a robot. You are being not-found by a search. The fix is not to outsmart an algorithm — it is to be legible to a parser and findable by a recruiter typing plain words into a box.

Understanding this changes what you optimise for. Tricks aimed at fooling a scoring system are pointless and often harmful. Making your document parse accurately and read in the employer's vocabulary is the entire job.

Five Claims That Are Wrong

Before the fixes, it is worth clearing out advice that will actively cost you.

Several of these myths appear in the wider list of things that quietly damage applications, covered in our rundown of the mistakes that get resumes rejected.

Formatting That Survives Parsing

This is where real failures happen. A parser reads a document as a sequence of text runs and infers structure from position and headings. Anything that disrupts that inference produces scrambled or missing data.

Things that genuinely break parsing:

• Multiple columns. The single most damaging choice. Many parsers read straight across the page, interleaving your left and right columns into unusable text. A skills sidebar next to your work history can turn both into nonsense.
• Tables. Cell contents are frequently read in an unpredictable order, or dropped. Using a table to lay out contact details or skills is a common and costly habit.
• Text boxes and floating shapes. Content inside them is often outside the main text flow entirely and simply does not get extracted.
• Document headers and footers. Many parsers skip these regions. Putting your name, phone, and email there means the record arrives with no contact information.
• Images and graphics containing text. Logos, icon-based contact lines, and graphic skill bars carry zero extractable text. A skill shown as a five-star rating is, to the system, not present.
• Image-based PDFs. Exports from Canva, Figma, Illustrator, or a scanner. The page looks like text and contains none.

Things that are fine despite the folklore: bold and italic text, bullet points, a single horizontal rule between sections, one restrained accent colour, standard fonts at any reasonable size, and a two-page document.

Section headings matter more than people expect. Parsers look for conventional labels to decide where each block of text belongs. Use "Work Experience," "Education," "Skills," "Certifications." Creative alternatives like "My Journey" or "What I Bring" can cause an entire section to be misfiled or ignored. This is one case where convention is worth more than personality — the personality belongs in the content, as covered in our guide on making a resume memorable without gimmicks.

Two structural details worth getting right: put every date in a consistent, unambiguous format such as "Mar 2022 – Jun 2024," since inconsistent dates are a leading cause of mangled employment histories; and keep each job's title, employer, and dates on the same line or immediately adjacent lines, because parsers associate them by proximity.

How Keyword Matching Actually Works

Recruiters search the database using terms from the role they are filling. Some systems also compute a similarity score between your text and the posting. Both mechanisms reward the same thing: your resume containing the words the employer uses.

The problem is that equivalent ideas have many names. A posting says "stakeholder management"; your resume says "worked closely with senior leadership." A human reads those as the same. A string search does not.

A practical process for closing that gap:

1. Read the posting and mark every concrete term — named tools, methodologies, certifications, and the recurring phrases in the requirements section. Terms repeated two or three times are the ones the recruiter is most likely to search.
2. Sort them into three groups: present in your resume already, true of you but worded differently, and not true of you.
3. Rewrite the second group using the posting's exact phrasing. This is the entire exercise, and it is honest — you are describing real work in the employer's vocabulary.
4. Include both forms where an acronym is involved. "Search Engine Optimization (SEO)" catches searches for either. Same for "Certified Public Accountant (CPA)" and similar.
5. Place terms where they are credible. A skill that appears in the skills list but never in the work history reads as thin to the human who eventually opens the file. Ideally each important term appears once in the skills section and once inside an achievement bullet.
6. Leave the third group alone. Do not add capabilities you lack to raise a score; the interview will find them.

Your professional summary is unusually valuable here, because it sits at the top of the extracted text and is a natural place for the target job title and two or three core competencies to appear without straining. Our guide on writing the opening summary covers the structure. The skills section carries similar weight, and the guide on deciding which skills belong explains how to keep it specific rather than a keyword dump.

Doing this by hand takes fifteen to twenty minutes per application, which is why most people skip it. Automating the comparison is the fastest route: paste your resume alongside the posting and the AI returns the specific missing terms and a rewritten version that includes them naturally.

Which File Format to Send

Both mainstream formats work when generated properly. The failure cases are specific:

• PDF is the safer default for most applications. It preserves layout exactly and every current ATS reads text-based PDFs. Export it from Word, Google Docs, or a builder — not from a design tool.
• DOCX is marginally more reliable across older and less common systems and is the right choice when a posting specifies it or when a recruiter asks for an editable copy.
• Never send .pages, .odt, an image file, or a link to a hosted document unless explicitly requested. A link in particular often cannot be followed by the intake process at all.
• Follow the posting's instruction whenever it gives one. That instruction usually reflects a known limitation of the system they are running.

The filename matters too, in a small way that costs nothing to get right: FirstName-LastName-Resume.pdf. It appears in the recruiter's file list and in the ATS record, and "resume_final_v3.pdf" is both anonymous and slightly careless.

How to Test Your Own Resume

You do not need specialist software to catch the majority of problems. Three checks, in increasing order of effort:

1. The copy-paste test. Open your PDF, select all, copy, and paste into a plain text editor. What you see is approximately what the parser sees. If sections are interleaved, dates have drifted away from their employers, or your contact line is missing, you have a parsing problem. If nothing pastes at all, you have an image-based PDF and need to rebuild it.
2. The read-aloud order check. Read that pasted text top to bottom. Does the sequence make sense on its own, with no visual layout to help? A parser has no layout — this is the version it stores.
3. A parser or scanner tool. Several services will show you the structured fields a system extracts from your file. Jobscan is the best-known for keyword gap analysis against a specific posting; most resume builders and job boards also expose some form of parse preview. FixResume approaches it from the rewriting side — it identifies the gaps and produces the corrected document rather than only reporting a score.

If the copy-paste test comes back scrambled, repairing a heavily formatted document is usually slower than rebuilding it. The free resume builder outputs a single-column, text-based layout that passes these checks by construction, which removes the whole category of problem in one step.

Once the document parses cleanly, everything else about resume quality still applies — the parser only decides whether you are findable, not whether you are persuasive. The broader diagnostic process is in our guide on fixing a resume that isn't getting callbacks, and if you are weighing whether AI tools handle this well, the honest assessment is in our look at using ChatGPT for resume work. Credit options for bulk tailoring are on the pricing page.