By Alan Price
The numbers paint a stark picture: one-third of job seekers are now using AI in their cover letters, and the challenge goes beyond polished applications. As many as a quarter of all job applications could be deepfakes or fraudulent by 2028, according to Gartner research.
Tenured talent acquisition leaders are witnessing this evolution first-hand. Candidates arrive with AI-polished resumes, rehearsed responses generated by ChatGPT, and interview answers that sound perfect on the surface. Although recruiters are getting more polished candidates, they're not necessarily prepared to actually do the job. Add to this the fact that many hiring decisions still rely heavily on gut instinct and personal chemistry, which research shows doesn't totally correlate to performance on the job.
This disconnect comes at a critical time. With AI adoption happening on both sides of the hiring equation, the fundamental evaluation process remains broken.
Million-dollar Missteps
Historically, interviews have been black boxes. A hiring manager walks into a room, asks questions, fills out a scorecard, and makes a recommendation. But what actually happened in that conversation? What competencies were genuinely assessed versus assumed? Without visibility into the actual interview content, organizations are making million-dollar hiring decisions based on incomplete information.
The first step in building a data-driven approach is to recognize that any AI tool used in this process is an enabler for the people making decisions—but no one should be outsourcing decisions to AI.
Deel started using Metaview, an AI tool which records video interviews, turns voice into text, and helps summarize the interview. Instead of chronological notes, recruiters now categorize every exchange into competency areas: functional skills, practices, behaviors, achievements, and motivations. This shift from narrative to structured data changed everything.
The biggest surprise? After analyzing the data, there were massive gaps between what was claimed to be evaluated and what was actually assessed. For example, critical thinking was one of the core competencies identified for customer success. Yet only a small percentage of those interviews actually specifically asked questions about critical thinking. Turns out, while this was set as a priority, interviews didn’t account for it.
Another concrete example is Deel’s sales department. Managers wanted experienced candidates who could "hit the ground running," but gut-check hiring led to overqualified five-year veterans who churned after 12 months. The data revealed that some of the best performers actually had two to three years' experience from startups with hard-to-sell products. They may not have had shiny CVs, but they had the resilience that was needed.
This required training for the team by giving them all of the data, and showing them that the candidates they had chosen based on particular pedigrees, such as from a brand-name organization, didn't always prove to be the best performers. The data showed them that they needed to interview more candidates and consider those with less experience.
The most eye-opening revelation was during the process of scoring interviewers, not just candidates. Deel tracked the complexity of their questions, their probing techniques, and how thoroughly they covered required competencies. Each interviewer was ranked based on their probing questions and the degree of complexity of those questions. Then, each interviewer received a score that then helped determine if this person is a reliable source of information.
The results were stark. Some interviewers consistently identified candidates who become high performers, while others didn't. Others chewed up as much as 70% of the interview talking and didn’t even get through all the core competency questions.
The shift to data-driven hiring doesn't require abandoning human judgment. It requires enhancing it with better information. Remember: It is an enabler for the people making decisions in the process. Hiring managers should always make the final decision. But with more data and more insights, they are able to look more holistically at what correlates to success. Here are some best practices to consider.
Structure a competency framework. Structured interviews have been around for years for a reason. When building a blueprint, look at five areas: skills, practices, behaviors, motivations, and achievements.
Implement interview deep dives. Deel uses frameworks like Star AR: situation, task, action, results, and AR is alternative action, alternative result. It's even more critical today to dive deeper into AR instead of scratching the surface. This helps a hiring manager understand if that person actually has the capability to do the job as opposed to presenting well because they got a leg up with AI.
Track interviewer effectiveness. Having conversational data provides insights into what was covered during the interview and the quality of the questions.
It's currently an arms race between candidates who are using AI to boost their chances and employers who have to integrate AI tech as well as conduct identity checks, verify employment and education, and even make sure virtual interviews aren't manipulated.
But the solution isn't to retreat to pure intuition. The technology exists to transform hiring from art to science. The companies that will win the talent war aren't those with the best gut instincts, they're the ones making decisions with the best data.
Alan Price is director of talent acquisition at Deel.