The ATS rejection myth: what software actually does to your CV

Somewhere between 2015 and 2020, every CV-advice article on the internet converged on the same statistic. You have seen it too.

"75% of CVs are rejected by an ATS before a human ever reads them."

It is on LinkedIn posts, in career coach newsletters, in the sales copy of every CV-optimisation tool sold on the internet. It is treated as common knowledge. It is not.

The number has no primary source. It appears to have originated from marketing content produced by Jobscan, a commercial ATS optimisation tool, and has been recycled uncritically ever since. When researchers have actually surveyed recruiters on whether their ATS auto-rejects CVs, the answers have come back in the other direction. A 2025 Enhancv study found 92% of recruiters said their ATS does not auto-reject anything. A November 2025 analysis called the 75% figure "a myth that has taken on a life of its own, driven largely by companies that sell CV optimisation services."

Which leaves a reasonable question: what does an ATS actually do?

What the software is, and what it is not

Workday, Taleo, Greenhouse, Lever, SmartRecruiters, iCIMS. These are the systems most companies are running. They are not sentient gatekeepers making hiring decisions. They are workflow and database tools.

What they do: receive an application, parse the content into structured fields (name, employer, job title, education, dates), store it in a searchable database, and present applications to recruiters in lists that can be ranked or filtered.

What they do not do, in most implementations: automatically discard applications without a human configuring that filter. Independently score a CV against some universal standard. Apply consistent logic across employers. Every ATS is configured differently by whoever sets it up.

A recruiter who has worked inside three of the main systems put it plainly on LinkedIn recently: "There is a massive misconception that applicant tracking systems auto-reject resumes. They don't. Recruiters and hiring managers are the ones who set up filters, and most don't set hard knockout criteria."

The practical consequence is subtler, and worse, than the myth. Your CV is not rejected. It sits in the database at a very low rank, essentially invisible, and the recruiter running the search never clicks past the first few pages to find it.

What the best study actually found

The most credible research on any of this is Harvard Business School's "Hidden Workers: Untapped Talent" (2021), by Joseph Fuller and Manjari Raman. They surveyed 8,000 workers and over 2,250 employers. It is the largest serious study on ATS effects ever conducted.

Their finding was not that ATS software is rejecting good CVs over formatting.

Their finding was that an estimated 27 million qualified workers in the US were being filtered out by automated systems, and the mechanism was minimum-qualification filters set too rigidly by employers. A role required a specific degree, or a continuous employment history, or an exact job title match. Candidates who could genuinely do the work were eliminated because they did not fit the filter's narrow shape.

88% of employers acknowledged the problem for high-skills roles. 94% acknowledged it for middle-skills. Most knew what was happening and said they could not afford to loosen the filters because doing so would drown the recruiter in volume.

That is a different problem from the one the optimisation industry sells a fix for. No amount of keyword-stuffing or formatting polish on a candidate's CV will change a filter that demands a specific degree the candidate does not have. The fix Harvard recommended was that employers change how they configure their systems, not that candidates buy CV-optimisation software.

Basic hygiene vs optimisation

Formatting does matter in one limited sense. ATS parsers do fail on certain things: text inside images, multi-column layouts, creative fonts that get garbled, contact information in headers and footers, non-standard date formats. A 2025 analysis of 1,000 rejected CVs by Edligo found these were the most common parsing-related failures.

All of these are avoidable without any commercial tool.

Use a clean, single-column layout. Use standard fonts. Put contact information in the body of the document, not in a header or footer. Use standard section headings: Experience, Education, Skills. Keep dates in a normal format. Do not hide key text inside graphics.

That is basic hygiene. It takes thirty minutes. It is not "optimisation." You do not need to pay anyone £150 for it.

The commercial step up from hygiene to optimisation, matching the exact keyword density against a specific job description, gaming match scores, using paid tools to verify ATS compatibility, is where evidence gets thin. Jobscan's own marketing material claims that users who achieve a 75%+ match rate receive more interviews. That data comes from Jobscan's own paying users with no control group. It is self-serving self-report. There is no randomised controlled trial, anywhere in the peer-reviewed literature, comparing outcomes for "ATS-optimised" CVs against plain ones. The evidence base for the entire optimisation industry is anecdotal and commercially motivated.

Keyword stuffing is now counterproductive

There is also a practical ceiling on the optimisation strategy. Modern ATS platforms have become better at detecting keyword stuffing, and newer AI-powered screening tools apply semantic analysis rather than raw keyword matching. A CV crammed with every term from the job description used to game the match score will now often be flagged as suspicious.

More importantly, if the CV gets past the software, it then has to be read by a human. A CV that has been written to hit keyword density targets reads like it has been written to hit keyword density targets. Recruiters notice. Up to 80% of hiring managers in recent surveys say they reject CVs that sound AI-generated or keyword-dense and lacking a genuine human voice.

You end up optimising yourself into exactly the pile you were trying to escape.

The commercial incentive problem

It is worth sitting with the shape of the industry that sells the fix.

The CV optimisation industry is substantial: Jobscan, Resume.io, Resumeworded, Enhancv, Reztune, and dozens of competitors. Typical services charge £50 to £200 to "optimise" a document against the ATS threat. Their entire business model depends on candidates believing the threat is real and severe.

The logic is elegant if you look at it coldly. Create fear about an invisible, automated process that candidates cannot see or verify. Sell protection against it. Because candidates never get feedback on why their application failed, "the ATS rejected you" is unfalsifiable. You cannot disprove it. You just paid someone to make it go away, and the next time you do not hear back, you cannot tell whether the optimisation worked or whether you were just not the best candidate.

This is not a moral judgement about the people in that industry. It is a structural observation. An entire commercial sector is built on a statistic with no source, selling protection against a problem whose severity is impossible to verify.

Why at senior levels none of this really matters

For most candidates reading this, the above is academic. Here is the part that matters.

Senior and executive hiring does not run through ATS at all.

Director, Head of, CFO, Managing Director, Partner. These roles are filled through headhunters, referrals, and proactive search. A retained executive search firm does not filter candidates through an ATS. They identify targets, make approaches, build shortlists through networks, and present candidates directly to hiring committees. When a senior candidate does apply to a posted role, their CV is typically reviewed by a senior recruiter or the hiring manager, not ranked by keyword match software.

LinkedIn is the effective database for senior profiles. Recruiters search LinkedIn directly, outside any ATS. This is particularly true in the spaces I work in most often: PE-backed finance, infrastructure investment, specialist commercial leadership. Those roles are almost universally filled through retained search or direct network, not through advertised ATS-screened listings.

At director level and above, the CV is not trying to pass a bot. It is trying to prepare a hiring manager for a conversation. The narrative, the evidence of commercial judgement, the quality of the career story, these are what matter. A keyword-optimised CV written to beat a system that does not apply at this level is, at best, a waste of money. At worst, it produces a CV that reads as though it was written by someone more interested in gaming software than in telling the truth about themselves.

What actually makes a difference, by level

Here is a more defensible framework for where to put your effort.

At graduate or high-volume application level. Clean formatting and reasonable keyword alignment are genuine hygiene. The ATS and the recruiter running keyword searches are both real, and both are looking at hundreds of applications. A well-written CV using natural language from the job description will not get a weak candidate hired, but it reduces the risk of a strong candidate being invisible.

At mid-level. The ATS is a sorting mechanism, not a gatekeeper. A well-written CV with a clear career narrative and quantified achievements, using language consistent with the job description, will do fine. Clarity beats optimisation here.

At senior and executive level. The CV is a conversation preparation document, not an application filter. What matters is narrative, commercial evidence, LinkedIn quality, and the referral or search channel you are going through. ATS optimisation is irrelevant.

The simpler verdict

Spend thirty minutes making sure your CV is cleanly formatted in a single column, with standard fonts, standard section headings, and all your text readable as text (not inside images).

Use natural language that matches the roles you are applying for, because that is good writing, not because an algorithm is watching.

Do not pay £150 for an optimisation service. Do not keyword-stuff. Do not restructure a strong executive CV to pass a bot that, at the level you are targeting, is not the decision-maker anyway.

The CV problem most people have is not that software rejected them. It is that the document did not earn a human reader's attention. That is a positioning problem, not a formatting problem. And positioning is not something a template or a parsing tool can fix.