Speed Hiring vs. Diversity
A Hiring Dilemma

Her team has managed talent acquisition for Amazon's India Stores, Retail Business Services, and its high-growth business units. Over several years, Kothari and her team helped build and scale some of Amazon India's most ambitious diversity hiring pipelines — programmes designed to recruit women returning from career breaks, persons with disabilities, LGBTQI+ candidates, and military veterans.

In 2023 and 2024, those programmes came under pressure from a different direction: the demand for speed. In 2025, Amazon.com removed a reference to "inclusion and diversity" in its annual report filed in February 2025. This got Shreya worried and thinking — was it time to deprioritize diversity initiatives and focus on speed in India as well?

Amazon India had the finest talent team that helped find and hire new people to work for the company. Diversity channels operate on a different timeline and logic. Programmes like Rekindle (for women returning after career breaks), WoW — Women of the World (a 70,000-member engineering community), Elevate HER, a PWD Train and Hire programme, LGBTQI+ hiring through Pride Circle, and military veteran hiring each required months or years to build.

For these pipelines, time-to-hire often runs 60 to 120 days, sometimes longer when upskilling is part of the process. These programmes were created to address specific issues in India's job market, such as the low number of women in the workforce and the lack of opportunities for people with disabilities. Shreya found that just posting job ads wasn't enough to attract the right candidates — finding diverse individuals was not as easy as it looked on paper.

Amazon India was growing fast, and it was getting harder to find the right people. At the same time, costs needed to be cut. The timeline to fill jobs had become a big deal, and top executives started checking on it every week. Meanwhile, the money, staff, and time available for diversity hiring — the three essential things needed to make it work — were all reduced. Although no one officially asked to stop hiring diverse candidates, the incentives now focused on speed. Requisitions earmarked for diversity channels were gradually moved back to the regular hiring process, and diversity hiring became less of a focus.

Amazon India's corporate hiring was spread across six levels — L2 through L8 — with the bulk of the volume concentrated at L4 and L5. This included software engineers, programme managers, category managers, and finance professionals who form the operational core of the business. L8 and above roles were handled in single digits per quarter and operated under an entirely different set of rules. The organisation ran two parallel recruiting tracks: a technology stream and a non-technology stream covering business, operations, and category roles — each with its own sourcing logic, assessment approach, and competitive landscape.

None of the diversity programmes at Amazon India came from headquarters. The global framework existed — goals, dashboards, representation targets — but the programmes themselves were built locally, because the problems were local. Women's workforce participation in India sits at around 24 percent. Employment of persons with disabilities in corporate India is, for practical purposes, near zero.

Standard recruiting tools — job boards, LinkedIn, campus postings — could not solve these problems because the candidates were not in the standard system to begin with. Every one of the diversity programmes followed a similar path: find a population the market ignores, then build the infrastructure to reach it.

Building talent pipelines of future employees took three types of investment. Shreya classified them as team, budget, and partnerships. Individually these things cost very little, but as a whole the cost was substantial. Pipelines were built through three years of work with NGOs such as Pride Circle, Dr. Reddy's Foundation, and the Skill Council for Persons with Disability, alongside campus work to grow women-focused programmes.

Shreya highlighted three types of evidence that the investment was working. The WoW programme had grown to 70,000 engaged women engineers — a proprietary talent community no competitor could reach. Crucially, the hiring bar stayed the same for diverse cohorts: the investment went into preparation, not into lowering the bar. Early data from Rekindle hires suggested they were staying longer than the market average for similar-level roles.

The downside: she had no hard numbers tying the investment to tangible business results. There was no definitive way to know whether Rekindle hires were promoted at the same rate as laterally hired peers, whether they were acquired and developed at the same cost as traditional hires, or — most importantly — how to attribute revenue to specific hires.

A 400-candidate pilot had already delivered a 12% reduction in cycle time — hard to match in a programme designed to be slower. Three individually manageable constraints — a reduced budget, reduced headcount, and targets that stayed the same or increased — proved collectively unsustainable. And because these pipelines are relationship-based rather than subscription-based, the breakdown happened at a glacial pace, visible in real time: sourcing volume plummeted, follow-up with partners slipped from days to weeks to months, and hiring managers began asking for the fastest candidate rather than waiting for a Rekindle or WoW hire.

Amazon had several mature programmes in place. Rekindle and WoW were highly efficient at converting people to hires, with predictable conversion rates and cadence. The PWD Train & Hire programme was maturing — 50+ hires from two pilot sites and actively scaling. The Pride Circle and veteran hiring programmes were still early.

All the diversity programmes were interconnected. Stopping one risked losing the entire talent pool — the 70,000 WoW engineers, the trained candidates, and the near-competition-free access Amazon had earned. Losing partnerships would mean starting from scratch to rebuild trust with organisations that have long memories: 18–20 months to re-establish a baseline, and it would take just one story travelling through tightly connected communities — "Amazon stopped their programme" — to undo it.

Rebuilding a same-scale pipeline from scratch would likely cost 3× as much as the current programme, and rebuilding partner progress would take at least 12–18 months and as many resources. Replacing recruiters would mean competing with Google, Microsoft, large conglomerates, and heavily funded startups on sign-on and base pay. If diversity hires really do stay longer, their departure and replacement would add costs of 50–100% of base compensation per year. Add rising regulatory exposure under the PWD Act and ESG reporting, and the cost of closing the gap later would far exceed the cost of running these programmes while they are still mature.

Shreya faced a tough choice: keep working on the diversity projects that had taken years to develop but were finally starting to show results, or switch to initiatives that could bring quick gains but might matter less in the long run. This was a critical moment, and the outcome of her decision could have a lasting impact.

In traditional hiring, strong candidates often have multiple offers from various employers. In diversity hiring, Amazon is one of only a few employers with a structured programme to reach a given group of underrepresented talent in the first place. So while in traditional hiring Amazon needed to "beat" other employers for a candidate, in diversity hiring Amazon first needed to "make it into the final round" of candidates the person even considers.