The Best Hires Are Equal Parts Art and Science

I’ll start with a confession. I make mistakes too. In 25 years of recruiting I have placed people who didn’t work out, and every time it happens I go back through the search looking for the moment I should have caught it. Sometimes I find one. Sometimes I don’t, and that second answer is the one nobody in this industry likes to say out loud.

I talk a lot about the invisible wounds people bring into hiring, the scars from past bad hires and bad bosses that nobody mentions until they collide. I carry a few of my own. A search that went wrong stays with you, and it should, because it changes what you pay attention to the next time.

Because a bad hire is so costly, companies respond to the risk in a very understandable way. They turn up the science knob. More instruments, more testing, more of a model behind the decision. Every year there’s a new tool promising to take the guesswork out of hiring, and now AI is promising to predict success before anyone has even picked up the phone.

I use plenty of these tools. They’re useful. But I’m honestly not sure all that instrumentation hires better people than the old-fashioned way. A strong hunch. Past performance. Someone I trust telling me this candidate is good. Hiring is a bet on a person, made with the best information available at the time, and I’ve come to believe the best hires are equal parts art and science. The trouble starts when a company leans too hard into either one.

Assessment Tools Are Legitimate, but Each One Only Measures What It Was Built to Measure

The two main scientific tools available right now are cognitive tests and personality assessments. Both are legitimate, and companies are right to use them. A cognitive test like the Wonderlic tells you whether a candidate clears a threshold. That’s what it was built for, and it does it well.

Personality assessments are trickier, because they generally aren’t designed to eliminate candidates from a process at all. They’re built for team cohesion and communication, and the providers say so plainly. That doesn’t stop companies from using them as a filter anyway, which is a little like using a thermometer to take someone’s blood pressure.

AI is heading somewhere useful here too. I think eventually a tool will take the profiles of the people a company has hired successfully, weigh a candidate against them, and flag the closest match. I don’t know that the tool exists yet, but I have a feeling it’s coming, and it will be an actual step forward. It will also still be a data point, not a decision.

There’s a second place I expect AI to earn its keep, and it’s one I find more hopeful. Versatility. I was at an event recently where the speaker, a chief human resources officer, had come up through accounting. People don’t have to follow linear paths, but a nonlinear path looks unpredictable on paper, and unpredictable is exactly what a screening filter is built to reject. Tools that can actually show transferable skills, that can recognize a utility player instead of flagging a resume gap, will help companies see candidates the current science screens out.

What none of these instruments can do is fix a fuzzy target. A tool can only screen against a standard, and in my experience the searches that go wrong usually go wrong before any tool gets involved, back when the requirements were never clearly pinned down in the first place. Vague on paper stays vague through every round of interviews, no matter how much testing you layer on top. This is why I spend so much time on requirements gathering, and why, when a seat is vacant and a client feels their back against the wall, I’ll still ask us both to pause and check the slate against the list we agreed on before we let ourselves get carried past it.

Even a Perfect Process Leaves a Sliver of Unpredictability, Because Life Changes on Both Sides After the Offer

Even when everything is done right, conditions change. On both sides. And they can change within 30 days of a start date.

Earlier this year I placed a candidate into a role with an agreement that she’d work from the office a couple of days a week. Sixty days in, she was a top performer and the team loved her. Then her personal situation changed and the move she’d planned was off the table. No assessment predicted that. No algorithm would have known what to do with it.

The company had a decision to make. Let a top performer go over a logistics requirement, or adjust the requirements around the person who was actually delivering. They adjusted the requirements, which I thought was really impressive. Not every company can do that. It takes a certain security to admit that the plan you wrote before you met the person matters less than the person, and the companies that can make that adjustment tend to be the same ones that keep their best people for years.

This is also why we stand behind every placement with a 90-day guarantee. There are times when all cylinders are firing, the due diligence is done, the candidate has been screened against a clear set of requirements, and you still get it wrong. I’d rather admit that upfront than pretend the science can promise an outcome it can’t.

The Art Is a Trained Instinct, and It Matters Most at the End of a Search, When Everyone Stops Using It

So what’s the art? People hear “gut feeling” and picture a coin flip. That’s not it.

I will say my gut has never steered me wrong, and at this point in my career that has become very useful. But my gut isn’t really about judging someone’s appearance, or even the things they say. It’s more the way a candidate interacts with me across a whole process, dozens of small moments that slowly draw a picture in my mind. When that picture gives me a reservation, I act on it. I won’t put a candidate through to a client when something doesn’t sit right, because we’re only as good as the people we place.

That kind of read is built the way any skill is built. Years of conversations, repetition, and a willingness to be wrong sometimes and learn from it.

The art earns its keep at the exact moment everyone stops using it, which is the end of a search. That’s when emotions run high. The candidate starts visualizing themselves in the seat. The hiring manager starts picturing this person on the team, joining them at lunch. And those final checkpoints, the reference that hasn’t been called, the requirement nobody re-confirmed, get easy to miss. That’s the moment to slow down, not speed up.

And the art keeps working after the start date, if you let it. I check in at 30, 60, and 90 days and ask both sides a version of the same question. Is this the job you thought you were getting? Is this the person you thought you were hiring? That’s alignment, and no instrument measures it, because it only exists once two real people are actually working together.

I’d love for every person we place to become a top performer. That’s the goal I hold, even knowing it isn’t a realistic bar. The science gets you a strong, well-vetted slate. The art gets you the rest of the way.

After 25 years, I’ve come to treat my gut as an instrument too. It’s the one that took the longest to calibrate.