Most Indian companies have a diversity hiring commitment somewhere — in a policy document, a leadership speech, or an annual report. Most of them also have a hiring process that systematically produces less diverse outcomes than their stated intentions, because the process was designed without diversity outcomes in mind and a policy statement doesn't change process design. This post focuses on the practical layer — where Indian hiring processes actually lose diverse candidates, what changes in sourcing, screening, and evaluation produce measurably different outcomes, and how to track progress without mistaking process activity for actual results. RecruitKar's AI candidate discovery matches candidates by actual skills and experience, not keyword patterns that correlate with specific educational or professional backgrounds — one of the structural sourcing changes that makes a real difference.
Why diversity hiring fails at the process level
Before fixing anything, it's worth being honest about where the failures actually occur — because the common interventions (unconscious bias training, diversity mission statements, diversity-focused job posting language) address symptoms more than causes.
The places Indian hiring processes most consistently lose diverse candidates:
Sourcing channels that reproduce existing demographics. If primary sourcing is through employee referrals from a team that's already homogeneous, the resulting candidate pool reflects the existing team's network. Referral sourcing is efficient; it's also structurally self-perpetuating. A company whose engineering team is 90% men sourcing primarily through referrals will generate an 80–90% male candidate pool regardless of what the job description says about valuing diversity.
Job descriptions with unnecessary requirements that disproportionately filter specific groups. Research on gender and job applications consistently finds that women are more likely than men to self-select out of roles where they don't meet 100% of listed requirements — even requirements that are genuinely "nice to have" rather than essential. An inflated requirements list that includes 6 items where only 3 are genuinely required doesn't just produce a worse candidate pool in general — it disproportionately loses certain candidate segments who apply the requirements more literally.
Degree and institution filters that correlate with socioeconomic background. "IIT/IIM preferred" or "top-tier college required" requirements filter heavily on family background and access to resources in early education, not on the job-relevant capabilities those institutions are being used as a proxy for. For roles where the actual skill is assessable directly, this filter excludes a significant population of capable candidates who didn't have access to the institutions used as a filter.
Unstructured interviews that reward communication style over competence. A candidate who articulates ideas fluently in English, uses the right corporate vocabulary, and comes from a cultural background that matches the interviewer's gets an unconscious premium in unstructured interviews. These correlate with specific educational, class, and regional backgrounds — not with job performance. Unstructured "just have a conversation" interviews are among the most reliably biased evaluation formats available, and they're the default in most Indian companies.
Post-offer attrition from candidates who encountered a different reality than promised. Diverse candidates who join on the strength of a stated commitment to inclusion and encounter a working environment that doesn't reflect that commitment exit quickly. This produces a revolving door where diversity numbers are maintained at the top of the pipeline but collapse at the retention layer — and where the company concludes that diverse candidates "don't stay," rather than examining what they encountered when they got there.
Sourcing: where the pipeline gap usually starts
Diversify beyond referrals
For roles where diverse candidate pools are a stated priority, actively source from channels that reach beyond your current team's network. This means:
- Targeted job board presence on platforms with broader demographic reach — Naukri's reach across regions and backgrounds is considerably broader than LinkedIn for most role types in India
- College and institution partnerships beyond the traditional shortlist — partnerships with women's engineering colleges, institutions in geographically diverse regions, and vocational training institutions for relevant technical roles
- Community sourcing in professional networks that are explicitly diverse — women in tech communities (She++, WiT India, AnitaB.org India chapters), LGBTQ+ professional networks, professional associations with diverse membership
- Re-entry programs for candidates returning from career breaks — a significant population of highly capable women in India have taken career breaks for caregiving reasons and are actively looking to return to professional roles; most companies don't actively source this pool
Remove unnecessary qualification filters before posting
Before each job description goes live, audit the requirements list with one question per item: if a candidate didn't have this specific credential or experience, would we genuinely disqualify them even if everything else were strong? Requirements that survive this test are genuine requirements. Requirements that don't should move to "nice to have" or be removed entirely.
Specific filters worth examining for each role:
- Degree requirements where the actual skill is directly assessable (coding ability, analytical reasoning, communication)
- Top-tier institution preferences where the preference is a proxy for a quality signal that could be assessed more directly
- Years-of-experience minimums that were set as a proxy for seniority when a more direct assessment of readiness would serve better
- Specific prior company or industry requirements that reflect hiring managers' familiarity rather than genuine role requirements
Use AI matching rather than keyword search as the primary sourcing layer
Keyword-based search reproduces the same candidate pool repeatedly — because it finds candidates whose profiles use the same terminology as the hiring team's search string, which tends to correlate with educational and professional backgrounds similar to whoever wrote the job description. Context-aware AI matching that searches for candidates whose actual described experience is relevant to the role — regardless of how they describe it — surfaces a broader candidate pool from which diverse candidates are less systematically excluded.
RecruitKar's AI candidate discovery is built on semantic matching rather than keyword matching, which is directly relevant here: a candidate who describes equivalent experience in different terminology or comes from a less conventional background is evaluated by what they did, not by whether their words match the search string.
Screening: where the pipeline shrinks
Apply consistent, documented criteria before reviewing applications
Define what a qualified candidate looks like — specifically and in writing — before reviewing any applications. Screening criteria defined after a recruiter has already begun forming impressions are susceptible to retrofitting: unconsciously adjusting what "qualified" means to match the candidates who seem appealing from early in the pile.
A written screening rubric — three to five genuine requirements, stated in advance, applied consistently to every application — produces more consistent shortlists across reviewers and more defensible decisions than undocumented individual judgment applied to each application independently.
Run a blind screening pilot for specific roles
Blind screening — reviewing resumes with name, photo, and sometimes educational institution removed for the initial screening pass — reduces the influence of demographic signals at the first filter. This doesn't work for all roles or all contexts, but for roles where implicit bias in screening is likely to be meaningful, a blind first pass followed by standard review of shortlisted candidates is a low-cost intervention with consistent evidence of effect.
Track shortlist demographics against applicant pool demographics
If 35% of applicants for a role are women but 10% of the shortlist is women, that's a signal worth examining — something in the screening process, the job description, or the sourcing channel is filtering women out at a rate disproportionate to their representation in the candidate pool. Without tracking this, the disproportion remains invisible. With it, the specific stage where diverse candidates are being lost becomes identifiable and addressable.
RecruitKar's visual ATS pipeline tracks source and stage data that makes this kind of analysis possible — comparing applicant volume by source against shortlist composition, and identifying which stages see the largest conversion drops.
Evaluation: where the pipeline collapses
Structured interviews with job-relevant criteria reduce bias more than any single intervention
The research evidence on this is unusually consistent: structured interviews — where every candidate is asked the same questions and responses are scored against pre-defined criteria — produce more accurate predictions of job performance and less demographic disparity in outcomes than unstructured interviews.
The mechanism is straightforward: an unstructured interview leaves significant room for an interviewer's impression — which is shaped by communication style, cultural familiarity, physical presentation, and other factors unrelated to job performance — to substitute for genuine evaluation of job-relevant competency. A structured rubric with specific criteria closes most of that gap by giving the interviewer a defined standard to evaluate against rather than an open-ended overall impression to form.
For companies committed to diversity outcomes, moving from unstructured to structured interviews is the highest-leverage process change available — higher than any awareness training or stated commitment.
Include multiple interviewers with different backgrounds where feasible
A panel that shares the same background, communication style, and educational experience will produce judgments that reward candidates who match that background, communication style, and educational experience. Deliberately constructing interview panels that include different perspectives — different seniority levels, different functions, different regional or educational backgrounds — doesn't guarantee unbiased evaluation, but it reduces the concentration of any single perspective's blind spots in the final assessment.
Evaluate each criterion independently before the debrief
In a panel debrief, the most senior or most vocal person's opinion tends to anchor everyone else's assessment before independent scores are compared. Requiring each interviewer to complete their evaluation independently and submit it before the group discussion begins — as covered in more depth in guidance on structured interview processes — produces genuinely independent input rather than false consensus.
Retention: where most diversity programs quietly fail
A company that has improved its diverse hiring rate but sees disproportionately high attrition among diverse employees hasn't improved its diversity outcomes — it's just accelerated a revolving door.
The most common retention failures for diverse employees in Indian companies:
Isolation without community. A single woman in an all-male engineering team, or a first-generation professional in a team dominated by people from similar upper-middle-class backgrounds, often faces a kind of ambient isolation that doesn't involve explicit discrimination but makes belonging genuinely difficult. Structured mentoring programs, employee resource groups (ERGs), and deliberate inclusion in informal networking are retention tools, not just morale activities.
Microaggressions that go unaddressed. Well-documented in Indian professional contexts: comments about communication accent, questions about whether someone "really" got their role on merit, exclusion from informal information networks that operate through personal WhatsApp groups. These erode belonging progressively in ways that don't appear in exit interviews as "discrimination" but accumulate into departure.
Promotion rates that diverge from hiring rates. If diverse candidates are hired at a rate that reflects stated commitments but promoted at a lower rate than comparable non-diverse employees, the inclusion is in the door, not in the organisation. Auditing promotion rates by demographic category alongside hiring rates reveals whether diversity investment is producing equitable career outcomes or primarily producing a diverse entry-level cohort that exits at the first promotion cycle.
Measuring what actually matters
Diversity hiring is one of the areas where organisations most easily mistake activity (training conducted, interviews completed, posts published) for outcomes (actual change in who gets hired, retained, and promoted).
Diversity hiring is one of the areas where organisations most easily mistake activity for outcomes.
The metrics worth tracking:
- Candidate pool diversity by stage: What percentage of applicants, shortlisted candidates, offer recipients, and joiners represent your diversity targets? Comparing these percentages reveals which stage filters most disproportionately.
- Offer acceptance rate by demographic segment: If diverse candidates are declining offers at higher rates than non-diverse candidates, the offer stage (compensation, role framing, or the impression formed during the interview process) is a problem.
- 90-day and 12-month retention by demographic segment: If diverse hires leave within the first year at higher rates, it's a retention and inclusion problem, not a hiring problem.
- Promotion rate parity: Are diverse employees being promoted at the same rate as their non-diverse peers with comparable performance ratings?
Tracking these numbers honestly and sharing them with leadership creates the accountability that makes diversity commitments operational rather than aspirational.
Frequently asked questions
Is diversity hiring legal in India? Can we explicitly seek to hire more women or underrepresented groups? Indian employment law prohibits discrimination on grounds of religion, race, caste, sex, and place of birth. Affirmative hiring targets that explicitly exclude non-target-group candidates from consideration would create legal exposure. However, companies can legally and legitimately: source more broadly from underrepresented communities, remove unnecessary barriers in the hiring process, and set internal diversity targets for applicant pool composition and hiring outcomes — without creating formal exclusionary quotas. Many Indian companies and MNCs with India operations run active diversity sourcing programs entirely within this framework.
What's the most impactful single change a company can make to improve diversity outcomes in hiring? Diversifying sourcing channels — specifically, reducing dependence on employee referrals as the primary sourcing mechanism and actively sourcing from channels that reach underrepresented communities. This addresses the pipeline problem at its source rather than trying to fix bias later in a process that started with a homogeneous applicant pool.
How do we handle hiring managers who resist diversity hiring initiatives? Address the underlying concern directly rather than around it. The most common resistance is "I want to hire the best person, not to fill a quota" — which is a legitimate concern applied to a straw man. The practical response: diversity sourcing expands the pool of qualified candidates you're considering; a better pool produces better hires. The point is not to lower the bar but to ensure the bar is being applied to a representative pool. Data from companies that have implemented structured sourcing diversification — showing improvement in hire quality alongside demographic diversity — is more persuasive than a values argument alone.
Should we disclose our diversity metrics publicly? Increasingly, yes — for companies that have them and want to signal genuine commitment rather than aspirational rhetoric. Disclosure creates accountability, demonstrates seriousness to candidates who are evaluating employers partly on inclusion, and tends to accelerate internal improvement by making the numbers visible. Companies that aren't yet ready to disclose because the numbers don't reflect their stated values might consider that the gap between the commitment and the data is the signal that process change is needed, not that disclosure should be deferred.
How do we reduce caste-based bias in hiring without it becoming explicit? Caste-based discrimination in employment is prohibited under Indian law and is a documented reality in Indian professional contexts. Process-level fixes that reduce caste-correlated bias without making caste explicit include: blind screening (removing names and educational institutions from the initial review), structured interview rubrics tied to job-relevant criteria (which reduce the influence of cultural familiarity and communication style on evaluation), and institution-diversified sourcing (reducing over-indexing on a small set of elite institutions that reflect historical access inequality rather than current potential). These are the same interventions that reduce gender and other demographic bias — the mechanism of reducing unstructured subjective judgment works across bias types simultaneously.
Build a more diverse hiring pipeline with RecruitKar. AI candidate discovery that matches on skills and experience, not keyword patterns — plus structured screening and multi-channel sourcing that reaches beyond your current referral network. AI candidate discovery — find candidates by what they've done, not where they went to school.