ATS in India: How Resume Screening Actually Works

An applicant tracking system does not reject your resume. It is a database with a form attached. It stores your application, tries to pull structured fields out of your file, and gives a recruiter a way to search and filter. It has no view about whether you are any good.

That matters because almost everything written about ATS assumes the opposite — that there is a machine scoring you out of 100 and binning you at some threshold. There isn't. The number most often quoted about resumes being discarded before a human sees them gets repeated everywhere and sourced nowhere; we have never found a primary study behind it, and neither has anyone who cites it.

What follows is what the software actually does, which systems Indian employers actually run — measured from our own scrape of 429 company career boards — and what really decides whether you get a call.

What the software actually does

When you upload a resume, the system runs a parser over it. The parser's job is narrow: find your name, your contact details, your employers, your job titles, your dates, your education, your skills, and write them into database fields. That is the whole trick. It is not reading for quality, tone or ambition.

Everything downstream depends on that extraction working. If the parser reads your two-column layout as one interleaved stream, your job titles end up attached to the wrong companies and your dates land in the wrong rows. Nobody rejects you for it. You simply become invisible to the search that a recruiter runs three days later, because the field they filtered on contains rubbish.

The one place software genuinely rejects you without a human is the application form itself. Employers can configure knockout questions — notice period, current location, work authorisation, a minimum years-of-experience field — and answers outside the accepted range can close the application immediately. That is a rule a person wrote, enforced at submission. It has nothing to do with your resume's formatting.

There is no single “the ATS”

Advice written as though one system exists is the root of most bad ATS guidance. In reality you are dealing with a handful of products that behave differently.

JobTether runs a scraper that reads postings directly from company career boards — the actual systems employers host their openings on, not aggregator listings. In our August 2026 snapshot of 6,342 live Indian postings across 429 companies, the pool was spread across Greenhouse, Lever, Workable, Ashby and SmartRecruiters — each linked to its own public API documentation. Lever, Workable and Ashby together carried 4,301 postings, 67.8% of the pool. Large enterprises commonly run Workday instead, which is notably less forgiving about layout than the rest.

You can see which one you are dealing with without any tooling: the system is visible in the link behind the “Apply” button on a company careers page. That is worth thirty seconds when a role matters, because it tells you how strict the parse is likely to be.

Naukri and LinkedIn are a separate category. They are job boards with their own search and ranking for recruiters, not the employer's applicant tracking system — although applications submitted through them usually land in one anyway. Applying on the company board puts your file in directly, in the format you uploaded, rather than through a profile that may reformat it on the way.

The other reason generic ATS advice fails is that the target moves between postings. We compared 158,842 pairs of live Indian job ads and found that two postings sharing a job title agree on a median of 11% of their stated requirements. Whatever the parser does with your file, the keywords it is matched against are rewritten by every employer.

The “ATS score” problem

This is the single most misleading idea in the category, so it is worth being blunt: no applicant tracking system gives a candidate a score. Greenhouse does not. Lever does not. Workable does not. Some systems rank applicants internally for the recruiter, but that ranking is private, employer-configured, and not a number you will ever see.

So every “ATS score” you have encountered — ours included — is a third-party tool applying its own rules and presenting the result as a percentage. That is why the same resume scores 72 on one checker and 94 on another. The disagreement is not a bug; the two tools are measuring different things, and neither has access to the system you are applying into.

This does not make scores useless. A good checker is a fast, specific formatting audit: it will catch the table you forgot, the missing section heading, the skills that appear nowhere in your experience. Treat it as a checklist that finds real problems, and ignore the number as a prediction. We go further into that in how to tell whether a score is real and how a score is actually calculated.

What actually filters most people out

Formatting decides whether you are readable. It is not what decides whether you are considered. That is almost always the stated requirement, applied by a person.

We measured this directly. In our analysis of 6,309 live Indian job descriptions, 71.9% stated a years-of-experience requirement, and among those the median bar was four years. Measured against the whole pool, 70.2% demanded at least one year and 32.4% demanded five or more. A recruiter filtering on that field will remove most applicants before any question of wording arises.

The practical consequence is a reordering of effort. Getting your resume parseable is a one-time job that takes an afternoon. After that, the returns come from applying to roles whose stated bar you are actually near, and from making the specific evidence for that role easy to find — not from chasing a higher number on a checker.

The formatting rules that genuinely matter

Nearly all of this reduces to one principle: never make the reading order ambiguous.

• One column. Parsers read top to bottom, left to right. Two columns get interleaved into nonsense — the most common serious failure, covered in does ATS read columns.
• No tables or text boxes. Same problem. Content inside them is frequently extracted out of order, or skipped.
• Nothing important in a header or footer. Contact details placed there are regularly dropped, which is how a resume arrives with no phone number.
• Text, not pictures of text. A design exported as a flat image parses to nothing at all.
• Conventional section headings.“Work Experience” maps cleanly. “Where I've Made an Impact” does not.
• A text-based PDF or .docx. Both are fine nearly everywhere; the detail is in PDF vs Word.

Two tricks that circulate and should be avoided outright: white text keyword stuffing, which parsers extract regardless of colour and recruiters read as dishonesty, and keyword lists bolted onto the end of a resume with no corresponding experience. Both survive the parse and fail the human.

Frequently asked

Everything else we have written about ATS

Thirty-four guides, grouped by the question you arrived with.