What 250 Interviews Taught Me About Judging Potential

Between December 2022 and May 2024, I conducted 250 candidate evaluations as an Admissions Ambassador for Rice University's MBA and PhD programs. That's 250 conversations — each between 20 and 40 minutes — where my job was to assess whether someone had what it took to succeed in a rigorous academic environment. Not whether they'd been successful before. Whether they would be.

That distinction matters more than most people realize. And the patterns I found across those 250 conversations have reshaped how I think about talent, readiness, and what "potential" actually looks like from the evaluator's side of the table.

The credential trap

The first thing you learn after about thirty interviews is that credentials are almost useless as predictive signals. I spoke with candidates from Goldman Sachs and McKinsey who struggled to articulate what they'd actually learned. I spoke with candidates from firms no one had heard of who could dissect a strategic problem with genuine clarity.

This isn't false modesty or contrarianism. It's an empirical observation. The correlation between resume pedigree and interview quality was, in my experience, essentially zero. The people who'd been told their whole careers that they were impressive sometimes had the hardest time demonstrating it — precisely because they'd never been asked to do anything beyond state their credentials and wait for the nod.

What does correlate with interview quality? Three things, consistently.

Signal one: specificity under pressure

The single strongest signal I found was the ability to be specific when the conversation got uncomfortable. Not rehearsed-specific — genuinely specific. The difference sounds like this:

"I led a cross-functional team that delivered 20% cost savings."

versus:

"The project was supposed to reduce costs, but what actually happened was that I misjudged how the procurement team would respond to the timeline. We lost three weeks. The saving ended up at about 12%, and honestly, 8% of that was from a supplier renegotiation that would have happened anyway."

The first answer is polished. The second answer is real. And the distance between them tells you almost everything you need to know about a candidate's relationship with their own experience. People who have genuinely processed what happened to them can be specific about it — including the parts that didn't go well. People who are performing a version of themselves can only be specific about the highlights.

After about a hundred interviews, I could usually tell within the first five minutes which type I was talking to. Not because of charisma or confidence — some of the best candidates were visibly nervous. But because specificity has a texture. It includes doubt, qualification, context, and the occasional "I'm not sure, but here's what I think happened." Polish, by contrast, is smooth all the way down.

Signal two: the ability to think about their own thinking

The second pattern was metacognition — the capacity to step back from what you did and explain why you thought what you thought at the time, and what you'd think differently now.

I'd often ask some version of: "Walk me through a decision you made that you're still not sure was right." The responses fell into three categories:

Category A — candidates who couldn't think of one. They'd been so trained to present everything as a success that the question itself confused them. This was the weakest response, and it was disturbingly common among otherwise impressive candidates.

Category B — candidates who could identify a decision and explain what went wrong, but only in terms of outcomes. "I should have done X because Y happened." This is better, but it's still backward-looking. It tells me you can identify mistakes with hindsight. Everyone can do that.

Category C — candidates who could explain not just what they decided and what happened, but what their reasoning framework was at the time, why it seemed right then, and what assumption or blind spot it rested on. This is the answer that signals genuine intellectual growth. It tells me you're not just learning from experience — you're learning from the structure of your own thinking.

Category C candidates were rare. Maybe 15% of the 250. And they came from every background imaginable — engineers, marketers, military officers, teachers. The variable wasn't experience. It was reflective capacity.

Signal three: genuine curiosity about the program

This one sounds obvious but isn't. Most candidates do their homework. They know the program rankings, the flagship courses, the notable faculty. That's table stakes, and it tells me nothing about fit.

What tells me something is when a candidate has thought about what they don't know yet — and can articulate why this specific program is where they want to figure it out. Not "Rice has a strong finance curriculum." Rather: "I've been doing pricing work for four years and I've realized I don't actually understand competitive dynamics the way I thought I did. I want to sit in Prof. Koka's competitive strategy course because my instinct is that the models I've been using are missing something about how rivals actually behave, and I don't know what that something is yet."

That's not a perfect answer. But it's a real answer. It tells me three things at once: the candidate has genuine domain experience, they've hit the limits of what experience alone can teach them, and they've done enough research to know which specific resource at this institution might help. That combination — self-awareness, intellectual humility, and specific curiosity — is what readiness actually looks like.

What I got wrong

I should be honest about my own learning curve. In my first thirty or so evaluations, I weighted communication skills too heavily. Fluent, articulate candidates got more favorable assessments from me than they deserved — and quieter, more halting candidates got less credit for insights that were genuinely original but poorly packaged.

I corrected for this over time, but the bias is real and I suspect it's endemic to the admissions process everywhere. We systematically overvalue the ability to sound smart and undervalue the ability to be smart. These are different skills. They sometimes coexist, but they don't have to, and in my sample of 250, they correlated less than you'd expect.

The practical implication for candidates: if you're not a natural communicator, don't try to become one in the weeks before your interview. Instead, go deeper into the specificity channel. Be more precise, more concrete, more willing to say "I don't know" when you don't know. That kind of directness reads as confidence to an experienced evaluator, even when the delivery is imperfect.

The deeper question

After 250 conversations, I'm left with a question I don't fully have an answer to: are we measuring the right thing?

The admissions interview, as currently practiced, is essentially a test of reflective self-narration. The people who do well are people who can tell a coherent, specific, self-aware story about who they are and why they're here. That's a valuable skill. But it's not the only valuable skill, and I'm not convinced it's the most important one for predicting success in a rigorous academic program.

Some of the most intellectually impressive people I've met in academic settings are terrible at self-narration. They think in systems, not stories. They can solve a problem beautifully but can't tell you why they approached it the way they did. Our current process screens these people out — or at least disadvantages them.

I don't have a solution to propose here. But I think anyone involved in candidate evaluation should sit with this discomfort rather than resolving it prematurely. We've optimized our admissions processes for a specific kind of intelligence — narrative intelligence — and we should at least be honest about what we're selecting for and what we might be missing.

These patterns — specificity under pressure, metacognitive capacity, and genuine curiosity — are not things you can fake in a 30-minute conversation. They're the residue of how you've actually engaged with your own experience over years. That's what makes them reliable signals, and it's also what makes them hard to game.