Candidate screening in India: why a fit score is a liability

A hiring manager screening resumes for a fast-growing retail chain assumes one of two things, usually without examining either: either India has a broad law like the US's employment-discrimination framework, so screening criteria just need to avoid a known list of protected categories the same way a US business would — or India has no such law at all, so screening on whatever criteria seem efficient carries no particular legal exposure beyond general fairness.

Neither assumption is accurate, and the gap between them is exactly where a hiring process quietly picks up risk nobody built in on purpose.

India does not have a single, comprehensive statute governing discrimination in private-sector hiring the way some other countries do. The Constitution's Articles 15 and 16, which prohibit discrimination on grounds including religion, race, caste, sex and place of birth, bind the State as an employer and apply to public-sector employment — they do not, on their own, create an enforceable right against a private employer's hiring decision.

What exists instead is a set of specific, narrower statutes, each protecting a particular ground: the Rights of Persons with Disabilities Act, 2016 prohibits direct and indirect discrimination against a person with a disability in employment, including denial of a role or of reasonable accommodation; the Transgender Persons (Protection of Rights) Act, 2019 prohibits discrimination against a transgender person in recruitment, employment and promotion, in both government and private entities; the Maternity Benefit Act, 1961 makes it unlawful to discharge or dismiss a woman during, or on account of, a period of maternity absence covered by the Act; and gender-based pay discrimination, formerly under the standalone Equal Remuneration Act, 1976, is now addressed within the Code on Wages, 2019, which the central government notified into force, alongside the other three labour codes, on 21 November 2025. A hiring process that screens in a way that touches any of these specific grounds is on genuinely different legal footing from one that screens on a ground none of these statutes names — and 'India has no anti-discrimination law' is true enough as a broad generalisation to be dangerous, because it invites ignoring the specific statutes that do exist.

Against that backdrop, a numeric fit score with a ranked candidate list is a specific kind of liability, independent of which protected ground might be at issue in any individual case. A score that blends multiple signals — experience, keyword match, an assessment result, sometimes something as loosely defined as culture fit — into one number is, by construction, difficult to unpack after the fact.

If a rejected candidate later raises a discrimination claim under one of the specific statutes above, or simply asks why they weren't selected, a composite score offers no way to show which factors actually drove the outcome, or to demonstrate that a protected characteristic wasn't quietly correlated with a factor that fed the score — an employment gap that tracks with a maternity leave, a location pattern that tracks with a disability-accessibility need, an assessment format that disadvantages a specific group without anyone intending it to. A ranked list built from that score compounds the problem, because ranking implies a level of precision and defensibility about small differences between candidates that a blended score rarely earns. None of this requires bad intent on the hiring team's part — a well-meaning scoring model can produce an indefensible outcome exactly because nobody can point to which input caused which score.

What holds up better, procedurally, is a screening record that keeps individual signals visible and separate rather than collapsing them into one number: what specific, job-relevant criteria were checked, what was found for each candidate against each criterion, who made the decision to advance or reject a candidate at each stage, and whether that decision was made by a person or a system.

A record like this can be reviewed after the fact and defended criterion by criterion — 'this candidate was moved forward because they met the stated experience threshold and the interview panel's scorecard reflected X' is a defensible sentence in a way that 'this candidate scored 71 out of 100' is not, because the first shows its reasoning and the second hides it. This is also, separately, a better hiring decision on its own terms, independent of any legal exposure — a hiring team that can see which specific factor moved a candidate forward is making a more informed decision than one trusting an opaque composite number to have weighed things correctly.

The practical risk sits in the same place whichever wrong assumption a business started with. A team that assumed broad anti-discrimination coverage and screened cautiously on obvious categories, while still building an opaque scoring system, remains exposed on the narrower statutes that do exist — disability, transgender status, maternity, and now gender pay equity under the Code on Wages — precisely because avoiding the obvious categories does not mean the scoring model didn't proxy for one of them anyway.

A team that assumed no relevant law exists at all, and screened however seemed efficient, is exposed on exactly the same specific statutes, just without having thought about it at all. Neither team's actual legal exposure changes based on which assumption they were operating under — only their preparedness to explain a decision if it's ever questioned does.

The more useful question for a hiring team to ask is not 'does a law cover this' as a binary gate, but 'could we explain this specific decision, criterion by criterion, to someone asking why' — a standard a transparent, criteria-based screening record can meet and an opaque composite score generally can't, regardless of which of India's specific hiring-discrimination statutes ultimately turns out to be relevant to a given case.

A recruitment platform that structurally refuses to let an automated system move a candidate to an adverse outcome without a named human decision, and that reports the individual factors behind a match rather than a blended score, is building toward that same standard as a matter of process — but the underlying discipline, keeping reasoning visible rather than compressed into one number, is worth building regardless of what tooling a hiring team uses. Whether a specific screening practice or a past decision creates actual legal exposure under any of these statutes is a question for an employment lawyer familiar with the current state of the labour codes, not something this article is positioned to answer for a specific case.