800,000 Applied, 286 Were Hired: What Bending Spoons' CEO Gets Right About Interviews

Last year, Bending Spoons received 800,000 job applications. It hired 286 people.

That's an acceptance rate of 0.04% — 100 times more selective than Harvard, more exclusive than the NASA astronaut corps, and harder to crack than Citadel's quant internship program. The Milan-based tech conglomerate, which owns Eventbrite, Vimeo, and AOL among a portfolio of digital brands, has made extreme selectivity a deliberate and advertised feature of how it operates.

On July 15, Fortune published an interview with CEO Luca Ferrari that explained why. Traditional interviews, Ferrari told the Wall Street Journal, are "basically completely useless." A run-of-the-mill interview, he said, is "like tossing a coin."

His company built something to replace it. And whether you ever apply to Bending Spoons or not, the argument behind their process has implications for how every serious candidate should think about interview preparation in 2026.

Why the Standard Interview Fails to Predict Anything

Ferrari's critique isn't new — it's well-supported by decades of organizational psychology research. The typical job interview, the kind conducted with open-ended questions like "Tell me about a challenge you overcame," is a poor predictor of job performance. Interviewers are susceptible to halo effects, they're inconsistent in how they weight answers, and the conversational format tends to reward the people who are best at performing well in interviews rather than best at the job.

The format that replaces it in academic research is structured: identical questions for every candidate, scored against predetermined criteria, weighted toward skills that are actually predictive of job performance. Most companies know this. Most companies don't do it anyway, because building and maintaining a structured process at scale is expensive and the incentives for interviewers to freelance are strong.

Bending Spoons built the expensive version.

What They Actually Measure

Out of 800,000 applicants, roughly 60,000 made it through initial screening. Those candidates were put through a battery of tests measuring three things: reasoning ability, domain expertise, and motivation and ownership.

The reasoning tests weren't job-specific. They were designed to measure how well candidates solve unfamiliar problems under time pressure — the kind of raw cognitive capacity that predicts performance across a wide range of roles. Domain expertise was layered on top: a candidate could be a strong reasoner with shallow domain knowledge, or a deep expert with slower processing. Both dimensions mattered. Neither compensated for the other.

Motivation was the third factor — specifically ambition, sense of ownership, and the drive to acquire new skills quickly. Bending Spoons is explicit that long-term potential weighs more heavily than current expertise. They're not trying to hire the best person at the job as it exists today; they're trying to hire the person who will compound fastest over the next several years.

From that pool of 60,000, approximately 3,300 reached the interview stage. Of those, less than 9% received an offer. At each stage, a dedicated internal "talent squad" — data scientists, software engineers, and AI researchers whose full-time job is evaluation — scored results against a hiring algorithm and tracked hired employees at four, eight, twelve, and twenty-four months. The goal: continuously refine which early signals actually predict which outcomes.

The 286 hires were the output of a system built to find signal in noise, not to fill roles quickly.

The Broader Implication for Candidates

Bending Spoons is an extreme case. Most companies won't run you through a multi-stage test battery before the first human conversation. But the problems the Bending Spoons process is solving for are real in most hiring environments — and the 2026 market has been pushing more companies in this direction.

The June 2026 jobs report came in at 57,000 new payrolls, roughly half the forecast, continuing a pattern of compressed job creation against a candidate pool swelled by AI-driven layoffs. Fewer openings and more applicants per role has concentrated scrutiny on the interview stages that do exist. Companies that used to hire on general competence and conversational rapport have sharpened their evaluation criteria — even if they haven't rebuilt their entire process the way Bending Spoons has.

The practical consequence: being good at performing well in a traditional interview is less sufficient than it used to be. What Ferrari is calling useless — the interview that rewards polish, that scores charm, that can't distinguish between someone who lived the answer and someone who memorized a good response — is the format that the most serious hiring processes have been moving away from. And when those processes get more rigorous, the gaps they expose are real ones.

What This Actually Means for How You Prepare

Stop optimizing for answers, start optimizing for thinking.

Bending Spoons is transparent about testing reasoning and judgment, not domain recall. The candidates who advance through their process are the ones who can work through an unfamiliar problem clearly — not the ones who've memorized the best answers to the most common questions. The signal they're collecting isn't "this person knows the right answer." It's "this person shows their reasoning under pressure, identifies when they're uncertain, and reaches a defensible conclusion anyway."

That's a different preparation target than most candidates aim at. The common interview prep pattern — gather a list of likely questions, write out polished answers, rehearse delivery — trains the ability to retrieve a pre-formed response under moderate pressure. It doesn't train the ability to reason through something you haven't seen before. If the company you're interviewing at is running structured assessments or algorithm-scored processes, the second skill is what gets you through.

Practice it specifically. Take unfamiliar problems (case questions, logic puzzles, ambiguous situations from fields outside your own) and work through them out loud, with a time limit. What you're building is the habit of externalizing your reasoning — showing the steps, naming your assumptions, saying "I'm not certain here, but the direction I'd go is..." That's what structured scorers are looking for.

Demonstrate learning speed, not just current expertise.

This matters beyond Bending Spoons. The thesis behind their weighting of long-term potential over current expertise reflects something that's quietly shifted in how high-performing companies think about hiring: in a landscape where AI tools change the cost structure of many tasks monthly, the person who adapts fastest compounds more value than the person who's most competent at the current version of the role.

In practice, this means showing your trajectory — not just what you know, but how quickly you moved from knowing less to knowing more. Interview examples that include the learning curve ("I didn't know X when I started, so I did Y, and within Z weeks I was able to...") score differently than static competence demonstrations. The former shows adaptability; the latter shows a snapshot.

The interview stage is still where it's decided.

Even in Bending Spoons' process — the one that screens 800,000 people down to 3,300 before human contact — the interview stage remained. Less than 9% of people who reached it got offers. The algorithm that follows the interview is informed by the conversation that preceded it.

The same is true everywhere. Even the most data-heavy hiring processes converge on a live conversation before closing. That conversation carries more weight precisely because you've already survived the filters that got you there — both sides have invested more, the comparison set is smaller, and the margin between the candidates in the final pool is tighter.

That's the stage where preparation quality shows up most visibly. The difference between a candidate who had their examples ready, who could answer the follow-up without losing the thread, who could discuss the actual outcome and not just the action they took — versus the candidate whose answers collapsed at the second question — is hard to manufacture on the spot.

Meeting Copilot's interview assistant is built for that stage. Before any interview, it assembles a briefing pack from your calendar event — the company, the role context, your prep material — so that what you've prepared is actually accessible when you need it, not buried under anxiety or browser tabs. During interviews where real-time AI assistance is permitted, the overlay surfaces your own preparation back to you at the moment it's relevant.

Ask what the process actually tests.

After the Bending Spoons story circulated, a predictable follow-on happened: candidates Googling the company found their process documentation and adjusted their preparation accordingly. You can do the same for any company you're targeting. Most companies that use structured assessments — HireVue, Pymetrics, Workday's cognitive tests, custom reasoning batteries — publish at minimum a general description of what they're measuring. Many post practice versions.

The candidates who arrive prepared for the specific format they're facing — not a generic interview, but the actual test or structured panel or case format the company uses — are operating with a meaningful advantage over the ones who prepped for a conversation and encountered something different.

What Ferrari Gets Right (and What He's Leaving Out)

The "useless" framing is deliberately provocative. What Ferrari actually built is an argument that most companies are measuring the wrong things. Traditional interviews are poor predictors not because interviewing as a format is irredeemable, but because most interviews are run without consistent criteria, weighted toward factors that don't predict job performance, and vulnerable to interviewer bias in ways that structured assessment minimizes.

What he's leaving out is that building the kind of rigorous process Bending Spoons uses requires a level of investment — a dedicated talent squad, years of outcome tracking, a large enough applicant volume to generate statistical power — that most organizations can't replicate. For companies that don't have 800,000 applicants to filter, the traditional interview's disadvantages are real but the structured alternative is out of reach.

Which is why the traditional interview isn't going anywhere. Most hiring decisions will still be made in conversations. What's changing is how much those conversations are surrounded by structured data — pre-interview assessments, algorithm-assisted scoring, post-interview reference checks that are actually conducted — that make the conversation one input among several rather than the only thing that matters.

For candidates, the practical implication is the same either way: prepare specifically, prepare for the format you'll actually face, and be ready to show reasoning under pressure rather than just recall under pressure. The companies that have moved farthest in Ferrari's direction are grading both.

---

Sources: Fortune: Bending Spoons hired 0.04% from 800,000 applications · Entrepreneur: Getting Hired Here Is Harder Than Getting Into Harvard · Metaintro: Bending Spoons Got 800,000 Applications