Real-World Case Studies
From the course AI for Recruiters (Advanced)
Three organizations that deployed advanced AI recruiting — and the specific tactics that drove their results.
Case Study 1: Unilever — AI Video Screening at Global Scale
Unilever adopted HireVue's AI video interview platform for early-stage graduate recruitment across 68 countries. Previously, human recruiters reviewed thousands of recorded video interviews, which was inconsistent and took weeks. The AI analyzed verbal content, vocabulary, and structured answer scoring (not facial expression analysis, which Unilever specifically excluded).
Results over three hiring cycles:
- Volume handled: AI screened 250,000+ video interviews that previously required 90,000+ hours of recruiter review
- Time-to-shortlist: Reduced from 4 months to 4 weeks
- Diversity outcomes: Percentage of hires from non-target universities increased from 40% to 55%, because the AI evaluated answer quality rather than prestige proxies
- Candidate experience: Candidate satisfaction scores for the screening stage increased from 3.1 to 4.2 out of 5
The controversy: Unilever faced public scrutiny over algorithmic transparency. Their response — publishing a full methodology document and allowing candidates to request human review — became a model for ethical AI hiring disclosure.
Case Study 2: Amazon (Cautionary) — When AI Bias Goes Unchecked
Amazon developed an internal AI resume screening tool from 2014 to 2017. The model was trained on resumes submitted over a 10-year period — a dataset dominated by male candidates, reflecting the company's historical gender imbalance in tech roles.
The outcome: the AI systematically downgraded resumes that included the word "women's" (as in "women's chess club") and penalized graduates of all-women's colleges. Amazon scrapped the tool in 2018 after internal audits revealed the bias.
Lessons for advanced recruiters:
- Training data reflects history, not aspiration — if your historical hires skew toward one demographic, AI trained on that data will perpetuate the skew
- Mandatory bias audits before deployment — do not wait for a journalist to find the problem
- Audit continuously, not just at launch — a model that passes a bias audit in Year 1 can drift as the workforce and language evolve
- Comply with NYC Local Law 144 — New York City now requires independent bias audits for AI tools used in hiring decisions, with results published annually
Case Study 3: LinkedIn — Talent Insights for Strategic Workforce Planning
A Fortune 500 financial services firm used LinkedIn Talent Insights to map their competitive talent landscape before launching a FinTech division. Traditional workforce planning relied on internal HR data only; the AI augmented it with external labor market intelligence.
What they built:
- An AI-generated heat map of where their target skills (Python, machine learning, regulatory compliance) were concentrated by geography
- A competitor talent flow analysis showing which firms people with their target profile had previously worked at and moved to
- A compensation model benchmarked against real offers (not survey data) for each skill combination
Decision made: they established their FinTech hub in Austin rather than NYC based on AI analysis showing 40% lower competition for ML talent and 28% lower average compensation for equivalent profiles.
- Time to decision: 3 weeks vs. a typical 6-month manual workforce planning study
- Cost savings: $2.3M in first-year comp savings attributed to location choice
Key Takeaways
- •Unilever's AI video screening processed 250,000 interviews and reduced time-to-shortlist from 4 months to 4 weeks
- •Amazon's scrapped AI tool shows that training on biased historical data perpetuates and amplifies that bias
- •NYC Local Law 144 requires annual published bias audits for AI hiring tools — proactive compliance is essential
- •LinkedIn Talent Insights enabled a Fortune 500 to make a $2.3M compensation-savings location decision in 3 weeks
Progress is kept in this browser only.