How to Get Your Resume Past the ATS (What Actually Filters You Out)

The short answer

Most applicant tracking systems do not automatically reject resumes. They parse your document into structured fields and let a recruiter search and filter those fields. You get filtered out when your resume does not contain the terms a recruiter searches for, or when bad formatting stops the parser reading your experience correctly. Fix the parsing, then fix the language.

If you are qualified and never hear back, the popular explanation is that a robot deleted your resume before a human saw it. That story is mostly wrong, and believing it sends you chasing the wrong fixes: invisible white keyword text, template gimmicks, and tools that promise to beat a filter that usually is not switched on. The real problem is more mundane and more fixable.

Where did the 75 percent claim come from?

It came from a sales pitch. The claim that 75 percent of resumes never reach a human traces to Preptel, a resume-optimization vendor, around 2012. Preptel sold software to beat the filter it was describing. The company shut down in August 2013 without ever publishing a study, dataset, or method behind the number.

One tell that it was never real: the figure drifts. You will see 70 percent, 75 percent, and 88 percent quoted for the same claim. A genuine statistic has one value and one method behind it. This one has several values and none, because each retelling copied the last rather than the source.

Recruiters who work inside these systems describe something different. Recruiter Jan Tegze puts the share of applications reviewed by a human at 90 to 95 percent or more. Greenhouse, one of the major platforms, does not auto-reject at all: every application reaches a reviewer, and recruiters work through structured scorecards and search instead.

What does an ATS actually do with your resume?

It parses and it indexes. When you upload a file, the system converts your formatted document into structured data: name, contact details, each employer, each job title, dates, education, and skills. Those fields populate a candidate record. Nothing is scored out of ten and nothing is thrown away.

The filtering happens afterwards, and a person does it. A recruiter with 400 applications searches the parsed records for the terms that matter to them, then reads what surfaces. So your resume is not competing against an algorithm. It is competing for a place in a search result, and then for thirty seconds of attention.

This is the part most advice gets backwards. Getting through is not about satisfying a machine. It is about being findable by a human using a machine.

So what actually filters you out?

Two things, and only one of them is technical.

• Vocabulary mismatch. If the posting says Kubernetes and your resume says container orchestration, you do not appear in the recruiter's search for Kubernetes. You were never rejected. You were never retrieved.
• Parsing failure. If your experience sits inside a text box, a table used for layout, a graphic, or the document header, the parser may capture it as nothing. Your record then looks empty in fields that matter, no matter how good the underlying experience is.

There is a third filter that is worth naming honestly, because it is the one candidates can do least about. Harvard Business School and Accenture surveyed employers for Hidden Workers: Untapped Talent in September 2021 and found that 88 percent of employers agreed that qualified, high-skills candidates are vetted out of the process because they do not match the exact criteria in the job description. Note what that says. The criteria are set by people. The software enforces a filter a hiring team configured, such as a required credential or a minimum years-of-experience threshold. That is a hiring-practice problem, not a robot problem, and it is why the honest version of this advice has limits.

How should you format a resume so it parses correctly?

Boring formatting parses best. None of this is about tricking anything; it is about making your experience machine-readable so a recruiter can find it.

• One column. Multi-column layouts can be read out of order, interleaving two jobs into nonsense.
• Contact details in the body, not the document header. Header and footer content is the single most common thing parsers drop.
• No tables used for layout, no text boxes, no experience inside images or icons.
• Standard section headings: Experience, Education, Skills. Clever labels like Where I Have Made an Impact do not map to a known field.
• Standard dates in a consistent format, because employment history is parsed as structured records with start and end values.
• A normal font and a .docx or text-based PDF. Export from a word processor rather than scanning or flattening to an image.

Once it parses, fix the language. Read the posting and use its nouns for the things you genuinely did. That is not keyword stuffing, which is transparent to any human who reads the result and does nothing for you once you surface. It is naming your own work the way the person searching for it names it. Which terms to pick, and how many, is the next section.

Which resume keywords carry the most weight?

Take them from the specific posting rather than a generic list of strong resume words. Two postings for the same title at different companies use different vocabulary, and each one is its own target. In rough order of what earns you a place in a recruiter's search:

• The literal job title from the posting. Jobscan's analysis of applications through its own tool reports a substantially higher interview rate when the resume title matches the posting. That is vendor data from a company selling resume optimization, so treat it as directional rather than measured.
• Hard skills and named tools, spelled exactly as the posting spells them. Systems do not reliably treat Adobe Creative Suite and Adobe Creative Cloud as the same thing, so matching the posting's wording is the safer bet.
• Certifications and licences written both ways, as in Search Engine Optimization (SEO), because some systems will not connect an acronym to its spelled-out form.
• Education phrasing and years-of-experience language, mirrored from the posting rather than paraphrased.
• Terms the posting repeats. Frequency in the ad is a reasonable proxy for what the hiring team actually cares about.

Placement matters as much as presence. Put the top required skills in a plain skills section where an exact-match search will find them, then prove each one inside an experience bullet with a number attached. Think of it as translation rather than invention: if you managed client relationships and the posting calls that account retention, use their words for your work. If you could not defend it in an interview, leave it off.

How many keywords do you actually need?

Fewer than you would guess. Published guidance clusters around ten to twenty-five targeted terms per tailored resume, and covering most of the posting's language rather than every word of it. Those thresholds come from vendor blogs rather than tested research, so treat them as rules of thumb.

Our own view: a resume that mirrors the ad word for word reads as copy-paste to the human who eventually opens it, and repeating a term beyond a few contextual mentions buys you nothing. The goal is being findable and then being credible, and those two pull in opposite directions past a point.

Who this is not for

If you are getting interviews but not offers, none of this is your bottleneck and formatting will not help. If you are applying to small companies using an inbox rather than a platform, there is no parser in the loop at all. And if you are being screened out on a hard requirement such as a licence, a degree, or work authorization, the filter is a policy decision and a better-formatted resume will not change it.

One honest limit on our own tools too: tailoring a resume improves how findable and relevant you are. It does not manufacture experience you do not have, and any product claiming otherwise is selling the same thing Preptel was.

If you want the mechanical part done faster, Mbason AI tailors your resume against a specific posting, and the job application tracker keeps the record of where you sent what, so you can tell which version is actually earning replies.

Do I need to match the job description word for word?

No, and it backfires. Use the posting's exact nouns for tools, skills, and titles you genuinely have, because those are the terms a recruiter searches. Copying whole phrases you cannot back up gets caught the moment a human reads it, and the interview conversation is where that falls apart.

Is a PDF or a Word document better for an ATS?

Either works if the text is real text. A PDF exported from a word processor parses fine. A PDF that is a scan or a flattened image contains no readable text, so the parser sees an empty document. If a job portal names a preferred format, follow it, since that is the one their configuration expects.

Do ATS-friendly resume templates actually work?

The plain ones do, but not because they are special. They work because they avoid columns, tables, text boxes, and header content, which are the things that break parsing. A template marketed as ATS-optimized offers no advantage over a clean single-column document you format yourself.

Does adding white keyword text help you get past the filter?

No. Hidden white text is visible in the parsed record a recruiter reads, so it reads as an attempt to game the process rather than as a qualification. It also does not solve the actual problem, which is that your genuine experience needs to be described in the words the recruiter is searching for.

How do you find the right keywords for a job application?

Read the posting and list the nouns: tools, systems, certifications, the job title, and any skill it names more than once. Those are your candidates. Then keep only the ones you can genuinely defend in an interview, and use the posting's exact spelling for each rather than your own preferred phrasing.

Should you repeat keywords multiple times on a resume?

A few contextual mentions is plenty, and repetition beyond that adds nothing. Search either finds a term or it does not; frequency is not a ranking factor you can exploit. What extra repetition does reliably is make the document read as padded to the person who opens it.

Do resume keyword scanners and match scores work?

They are useful for spotting terms from the posting you have genuinely missed, and they are not a measure of whether you will get an interview. The score reflects overlap with the job text, not your fit for the role, so chasing a higher number past the obvious gaps is wasted effort.