How ATS Software Works, Step by Step

Here's how ATS software works: it reads your resume file, pulls the text out using optical character recognition and natural language processing, tags pieces of that text as "entities" — a company name, a date range, a degree, a skill — and drops each tagged entity into a searchable database field. It does not, by default, "score" or reject you on its own; a human recruiter searches and filters that database using keywords, date ranges, and yes/no knockout questions they set up themselves. According to Jobscan's manual review of Fortune 500 career pages, 97.8% of large employers run a detectable ATS of this kind, and the same parsing logic sits behind Naukri, LinkedIn, and most enterprise hiring systems used by large Indian recruiters. What breaks candidates is not a mysterious rejection algorithm — it is bad parsing, a keyword mismatch, or getting buried under volume.

What an ATS Actually Is

An applicant tracking system is recruiting software that stores every application a company receives, in one searchable place, instead of a folder of email attachments. Recruiters at Indian firms hiring at scale — think TCS or Infosys running a campus drive that pulls in tens of thousands of applications in a single cycle — could not physically open every file. The ATS lets them query the pool instead: "show me every candidate with 'SQL' and 'Power BI' in their resume, based in Bangalore, willing to relocate." Globally, the most common platforms by market share, based on Jobscan's analysis of more than a million resume scans across 12,820 companies, are Greenhouse (19.3%), Lever (16.6%), Workday (15.9%), and iCIMS (15.3%). In India, large enterprises more often run Workday, SAP SuccessFactors, or homegrown platforms like Darwinbox, while portals like Naukri run their own parsing layer for the resumes candidates upload directly.

Step One: Parsing — Turning a Document Into Data

The moment you upload a resume, the ATS does not "read" it the way a person does. It runs the file through a parser that extracts raw text, then applies Natural Language Processing techniques — specifically a method called Named Entity Recognition — to identify which chunks of text are which kind of thing: a person's name, a job title, a company, a date, a degree, a skill. This is the step where formatting choices matter enormously.

Why Layout Breaks This Step

A single-column resume with standard headings gives the parser a clean, linear stream of text to tag. A two-column layout, a table, or text inside a header or footer often gets read out of order or skipped entirely, because many parsers process the main body text flow and nothing else. The result is a candidate record with blank fields — no dates, no company names — even though the actual resume file looks perfectly fine to a human. For the exact formatting choices that avoid this, see our ATS friendly resume template.

Step Two: Indexing — Making the Data Searchable

Once your resume is parsed into fields, the system does not re-scan the original file every time a recruiter searches. It builds what is effectively a word-to-candidate map — an inverted index — so that when a recruiter searches "Python", the system instantly returns every candidate whose parsed record contains that word, rather than re-reading thousands of PDFs on demand. This is also how Boolean search works inside these systems: a recruiter can search "Python AND SQL NOT internship" to combine, include, or exclude criteria in one query.

Step Three: Filtering and Knockout Questions

This is the step most job seekers misunderstand. The ATS itself does not decide you are unqualified. A human recruiter or hiring manager configures rules in advance — commonly called knockout questions — such as "Do you have a minimum of 1 year of experience in customer support?" or "Are you willing to work rotational shifts?" If you answer in a way that fails the rule, the system moves your application to a rejected folder automatically, but the rule itself was set by a person, not invented by the software. A widely repeated claim online is that "75% of resumes are rejected by ATS" before a human ever sees them. This figure traces back to a 2012 marketing claim by a company called Preptel that published no methodology, sample size, or survey behind it — it is not a confirmed statistic, and we are treating it as an unverified industry claim rather than fact. A 2025 Enhancv survey of 25 recruiters across industries found that around 92% do not configure their systems to auto-reject resumes based on content or formatting at all; most auto-rejection is reserved for hard mismatches like work authorization or a missing mandatory certification.

OCR, and Why Scanned Resumes Fail Immediately

If your resume is a scanned image or a photograph saved as a PDF, the parser has to fall back on Optical Character Recognition to even attempt reading it, and OCR is far less reliable than reading text that was typed directly into the file. A resume exported straight from Word or Google Docs as a text-based PDF avoids this problem entirely, because the underlying text is already selectable and machine-readable — you can test this yourself by trying to highlight text in your own PDF before you submit it. If you cannot select individual words with your cursor, the file is almost certainly image-based and will parse badly.

Step Four: Ranking, and Why It Is Not Automatic Scoring

Some modern ATS platforms do offer a ranking or "match score" feature that estimates how closely a resume's parsed content aligns with a job description, and some also let interview panel members submit standardised scorecards that the system aggregates into a composite rating later in the process. But the baseline behaviour of an ATS is filtering and sorting, not independently judging candidate quality. If your resume seems to "score low," in most cases what actually happened is one of three things: your formatting caused a parsing failure (fields came out blank or garbled), your wording did not contain the terms the recruiter searched for, or your application is sitting correctly in the system but far down a list of hundreds of others for that posting. Understanding what counts as a good ATS score is more about matching a specific job description than hitting some universal number.

Where Human Judgement Re-Enters the Process

Once a resume clears parsing and shows up in a recruiter's filtered search, a person is reading it, not software. In larger Indian hiring processes — a TCS or Infosys campus drive, for instance — recruiters often work through shortlists in the ATS interface itself, adding notes, tagging candidates for specific interview panels, and, in more mature systems, collecting structured scorecards from each interviewer that the platform totals into a combined rating. That combined rating is a human-judgement aggregation tool, not the same as the earlier parsing or keyword-filtering steps. Confusing the two is where a lot of "the ATS rejected me" thinking goes wrong: by the scorecard stage, your resume already made it well past the software.

What people assume	What actually happens
"The ATS reads my resume like a human would"	It extracts text and tags entities; layout errors cause blank or scrambled fields
"The ATS auto-rejects most resumes"	Auto-rejection rules are set by a human recruiter for specific hard criteria, not by the software's own judgment
"There's one universal ATS score everyone is judged against"	Matching is calculated against the specific job description a recruiter is filling, so the same resume scores differently for different roles
"A fancy visual resume proves I'm detail-oriented"	Visual layouts (columns, graphics, icons) are the leading cause of parsing failures across ATS platforms
"Once I'm in the system, nothing else matters"	Recruiters still actively search and filter using keywords, so wording continues to matter after you apply

How This Plays Out on Naukri Specifically

Naukri works slightly differently from a company's own ATS, because you are not applying to one job posting inside a closed system — you are maintaining a searchable profile that many recruiters query directly. Naukri does publish a free ATS-style resume checker that scores formatting and keyword match against a job description, but the separate "resume score" or "profile completeness" percentage you see on your dashboard measures how many profile fields you have filled in, not how well your resume would parse inside a company's actual ATS. We cover this distinction in detail in how Naukri's resume tools actually work, because conflating the two is one of the more common and costly misunderstandings among Indian job seekers.

Where This Matters Most in the Indian Hiring Cycle

Volume is the real bottleneck in Indian hiring, and it is why the ATS exists in the first place. A single off-campus drive or a popular listing on Naukri for a generic role — customer support, business development, entry-level software roles — can draw several hundred to a few thousand applications within days. No recruiting team reviews that volume manually. The ATS lets a two- or three-person recruiting team at a mid-size company do what would otherwise need a much larger team: narrow a few thousand applications down to a shortlist of thirty or forty worth reading closely, using search terms tied directly to the job description. That mechanical reality is also why generic, one-size-fits-all resumes underperform — they are less likely to contain the specific terms that surface in a narrow, well-targeted search.

What to Actually Do With This Information

Once you know an ATS is a search-and-filter tool rather than a judge, the practical response changes. Formatting has to be clean enough to parse without errors — that is a one-time fix, not an ongoing battle. After that, the real, repeatable work is making sure your resume's wording contains the specific terms a recruiter for that specific role is likely to search for, which changes job posting to job posting. That is a wording and research task, not a trick to "beat" software. Three things to do this week: run your current resume through a text-editor copy-paste test to confirm it parses cleanly, rewrite your Skills section to mirror the exact phrasing used in two or three real job postings you are targeting, and stop reusing one generic resume for every application if the roles differ even slightly in required skills.