Search for advice on applicant tracking systems and you will find a genre of tips built on a single assumption: that a machine reads your resume, scores it, and throws most applications away before any person sees them. Work backwards from that assumption and you get the familiar advice. Strip the formatting. Match the keywords. Feed the robot what it wants.
The assumption is doing a lot of work, and it is mostly wrong.
What an applicant tracking system is for
An ATS is a database with a workflow attached. Its main job is administrative: take applications from several sources, put them in one place, keep a record of who applied to what and when, track each candidate through stages, and give recruiters a way to search and filter. Hiring at scale without one is a spreadsheet problem nobody wants.
To do that, the system has to turn your document into structured fields. It parses the file, extracts entities like employers, titles, dates, skills and education, and stores them as data. Then a recruiter can search and sort, the same way you would filter a spreadsheet.
Two things follow from that, and they are the only two that reliably matter.
Parsing can fail, and failure is invisible to you
If the parser cannot read your document, your record arrives with empty or scrambled fields. Not rejected, just thin. A candidate whose employment history landed in the wrong field does not show up in the searches a recruiter runs. This is the real formatting risk, and it is why plain structure beats visual design in a document whose first reader is a parser.
Ranking is not rejection
Most systems can score how closely an application matches a posting. That score orders a list. Ordering a list is not the same as removing anyone from it. The candidate ranked 180th out of 400 has not been rejected by software. They are simply far enough down a list that a recruiter working from the top may never reach them.
That distinction matters because it changes what you should do. You cannot argue with a machine that discarded you. You can improve where you land in an ordering, and you can apply where the ordering is shorter.
The knockout questions do the actual filtering
Where automatic elimination genuinely happens, it is usually not the resume parser. It is the application form. Work authorization, willingness to relocate, required licence or certification, minimum years in a specific field: these are configured by the employer as explicit rules, and answering outside the allowed range can remove an application without human review.
This is worth knowing because it is the one place where a few seconds of care changes the outcome. People rush the form and answer the resume carefully, when the form is the part with the hard gate on it.
Screening tools are regulated, which shapes how employers use them
Automated selection is not an unregulated free-for-all. Under the Uniform Guidelines on Employee Selection Procedures, a selection rate for any race, sex or ethnic group that is less than four-fifths, or eighty percent, of the rate for the highest-scoring group is generally regarded by federal enforcement agencies as evidence of adverse impact (29 CFR 1607.4(D)).
The practical consequence for a candidate is indirect but real. An employer that lets software silently discard most applicants on opaque criteria is building an evidence problem for itself. Large employers know this, which is one reason blanket automatic rejection is less common than the folklore suggests.
Where the variance actually lives
The largest resume experiment ever run sent more than 83,000 fictitious applications with randomized characteristics to jobs at 108 of the largest United States employers. It was designed to measure discrimination, and it found some. The finding relevant here is a different one: contact rates varied enormously between companies, and those company-specific differences were persistent over time and across locations, correlating with things like recruiting centralization and firm profitability (Kline, Rose and Walters, Quarterly Journal of Economics, 2022).
Read that as a job seeker and the implication is uncomfortable. A large share of what determines whether you hear back is a property of the employer you applied to, not of the document you sent. Two identical applications to two large firms can have very different odds, and no amount of formatting closes that gap.
That is not a reason to send a worse resume. It is a reason to stop treating the resume as the only variable, and to spend some of that effort choosing where to apply.
What to do instead
Write the document so a parser can read it and a person wants to. Standard section headings, real dates, one column, no text baked into images or headers. Our ATS guide covers the structural rules in detail.
Take the language from the posting itself rather than a generic keyword list, because the ranking compares your application to that specific job description and nothing else. That is what keyword optimization does, and it is also why stuffing a skills section with terms you cannot discuss in an interview is a bad trade.
Check the match before you send rather than after silence, which is what the resume score is for. Then answer the application form slowly.
For how the pieces fit together, see how it works and the plans page.
The useful version of the advice
Make the document machine-readable, because parsing failures are real and silent. Mirror the posting, because that is what the ranking compares against. Answer the knockout questions carefully, because that is where automatic elimination actually lives. Then accept that a meaningful part of the outcome was decided by which employer you applied to, and let that shape where you spend your next hour.
References
- Uniform Guidelines on Employee Selection Procedures, 29 CFR 1607.4(D), adverse impact and the four-fifths rule. govinfo.gov
- Kline, P., Rose, E. K., and Walters, C. R. (2022). Systemic Discrimination Among Large U.S. Employers. Quarterly Journal of Economics, 137(4), 1963 to 2036. nber.org/papers/w29053
Disclosure: This article is published by MyJobsSearch and reflects our analysis and commentary on career strategy and job search tactics. It is for informational purposes only and does not constitute professional career, legal, or financial advice. It does not guarantee employment outcomes, interview offers, or job placements. Career results depend on many factors specific to your situation, industry, and geography. Consult a career coach or qualified professional for personalized guidance.
| Measure | Result | What it tells a job seeker |
|---|---|---|
| Contact gap, distinctively Black vs white names | 2.1 points | Measured discrimination exists and is the study's subject |
| Between-company variation in that racial gap | 1.9 points | Employers differ from each other almost as much as the average gap itself |
| Between-company variation in gender contact gaps | 2.7 points | Some firms favor men, others favor women, despite no significant average gap |
| Average male vs female contact gap | not significant | Aggregate figures hide the firm-level differences that matter |
The candidate ranked 180th out of 400 has not been rejected by software. They are simply far enough down a list that a recruiter working from the top may never reach them.
MyJobsSearch