Latest / Hiring & Firing with Fexingo: Recruitment, Performance, and Workforce Decisions / How Pre-Hire Assessments Predict Performance Better Than Resumes
Transcript
- Lucas: A few months ago I was talking to a friend who runs talent acquisition at a mid-sized tech company. They were spending about 40 hours per hire just on resume screening and interviews. Luna: Forty hours? That's a full workweek per person. Lucas: Exactly. And despite all that time, their mis-hire rate was sitting around 30 percent. So they decided to try something different: a 15-minute online work sample test before the first interview. Within six months, their mis-hire rate dropped to 18 percent. Luna: That's a 40 percent improvement. And it's not just an anecdote — there's research backing this up. Lucas: Right. The academic meta-analyses have been saying for decades that structured assessments — especially work samples and cognitive ability tests — are far better predictors of job performance than unstructured interviews or resume reviews. But most companies still lead with the resume. Luna: Why do you think that is? Is it just inertia? Lucas: Partly. But I also think resumes give hiring managers a comforting illusion of control. You see a degree from a good school, a few impressive job titles, and you feel like you've learned something. The problem is, a resume tells you almost nothing about how someone actually works — how they solve problems, collaborate, handle ambiguity. Luna: There's also the bias issue. Resumes are full of signals that correlate with privilege — where you grew up, who you know, which internships you could afford to take. Lucas: That's exactly the argument Google made when they started Project Oxygen back in the early 2010s. They analyzed thousands of hiring decisions and found that the best predictors of future performance were structured behavioral interviews and work sample tests — not GPAs, not school prestige, not even previous job titles. Luna: And they actually changed their hiring process based on that data. That's rare. Lucas: It is. Most companies collect data but don't act on it. Google did. They reduced their interview loops, added more structured scoring, and started using these assessment tools more systematically. Their internal reports showed a 24 percent reduction in mis-hires within two years. Luna: Okay, so what does a good pre-hire assessment actually look like? Is it a multiple-choice cognitive test? A job simulation? Lucas: It depends on the role, but the best ones are what researchers call 'high-fidelity' — they mimic the actual work. For a software engineer, that might be a coding challenge. For a marketing manager, it might be writing a campaign brief from a hypothetical scenario. The key is that the test is designed to measure skills directly relevant to the job, not general intelligence or personality traits. Luna: Although, there is evidence that general cognitive ability is still a strong predictor, especially for complex roles. Lucas: That's true. The meta-analysis by Schmidt and Hunter in 1998 found that cognitive ability tests had a validity coefficient of 0.65 for job performance — far higher than unstructured interviews at 0.20. But work samples were even better, at 0.54. And when you combine both, you get the best prediction. Luna: So why isn't everyone doing this? Cost? Time? Lucas: Cost is a factor. Developing a validated assessment is not cheap — you need industrial psychologists, pilot testing, and ongoing validation. A good assessment for a single role can cost tens of thousands of dollars. But for companies hiring hundreds of people in the same role, the ROI is enormous. Lower turnover, higher productivity, less wasted training. Luna: There's also the candidate experience concern. A poorly designed assessment can feel like a barrier, especially if it's too long or feels irrelevant. Lucas: That's a real risk. If you make candidates jump through a 45-minute test that doesn't seem connected to the job, you'll lose good people. But a well-designed 10-minute work sample? Candidates often report that it gives them a better sense of the role and makes them more excited about the opportunity. Luna: I've also seen companies use assessments too early in the funnel, before they've even done a basic screen. That can backfire because you're testing people who are clearly unqualified anyway. Lucas: Good point. The best practice is to use assessments after a brief phone screen, but before the full interview loop. That way you're not wasting anyone's time, but you're still getting a predictive signal early on. Luna: One company that does this really well is a financial services firm I worked with. They use a 20-minute data interpretation test for all analyst candidates. It's not about whether you know Excel formulas — it's about whether you can spot trends in a dataset and articulate what they mean. Lucas: And how did that change their outcomes? Luna: Their six-month attrition dropped by 35 percent, and managers reported that new hires were fully productive two weeks faster on average. Lucas: That's exactly the kind of data point that should make every hiring leader pay attention. But there's a cautionary tale here too. Some companies have gone overboard. I remember a case where a retail chain started using a personality test that filtered out candidates who scored high on 'openness' — because they thought those people would get bored. But openness is correlated with creativity and adaptability, which were actually valuable for customer service. Luna: So they were screening out some of their best potential employees. Lucas: Exactly. And that's the danger of using off-the-shelf assessments without validating them against your own performance data. The only way to know if an assessment works is to track how scores correlate with actual on-the-job outcomes over time. Luna: It sounds like the technology is there, but the implementation is tricky. You need the data, the statistical literacy, and the willingness to change your process. Lucas: And that's where most companies stumble. They buy a tool, roll it out, and then never look at the data to see if it's actually helping. But the ones that do — the Googles, the financial services firms — those are the ones that get a real competitive advantage in talent. Luna: If this conversation has given you some practical ideas for your own hiring process — and I hope it has — we want to mention that the reason we can keep these episodes ad-free is listener support. If you've found value in what we're doing, you can help sustain it at buy me a coffee dot com slash fexingo. Lucas: Yeah, that's exactly it. We're not asking for a subscription or anything — just that if today's episode saved you some time or made you think differently, tossing a coffee's worth into the kitty goes a long way toward keeping this independent. Luna: Alright, back to the assessments. Lucas, what's one concrete step a hiring manager could take tomorrow to start moving in this direction? Lucas: Take one role — just one — and design a 10-minute work sample that mirrors the most critical task of that job. For example, if you're hiring a customer support rep, write a sample ticket and ask the candidate to draft a response. Score it blind, against a simple rubric. Then track how those scores correlate with performance after 90 days. Luna: That's low-cost, low-risk. And you'll learn a lot from just that one experiment. Lucas: Exactly. You might find that the assessment predicts performance better than the entire interview loop. And if it does, you've got a compelling argument for scaling it up. Luna: I love that. It's a small bet with potentially huge returns. Lucas: That's the thing about hiring — small changes in process can lead to outsized improvements in outcomes. But you have to be willing to test your assumptions. Luna: And to look at the data, not just trust your gut. Lucas: Right. So next time you're about to spend 40 hours screening resumes, ask yourself: could a 10-minute work sample tell me more?