Crucial Hire
The Behaviors AI Cannot Score
Two candidates score 95 percent. One gets cut in round two and one gets the offer. Nothing on either resume explains why.
By Zach Nold, Chief Talent Officer at Crucial Hire, an HR-only executive search firm based in Chicago, Illinois, placing HR leadership nationwide.
The short answer
Automated resume screening measures what a candidate has done. It cannot measure how they did it, and how they did it is what predicts whether they succeed in your seat.
Four things no keyword sorter can see: how someone handles being wrong, how they treat a junior person when nothing is at stake, whether the people they hired got promoted or turned over, and whether operations actually trusted them. All four are knowable. None of them are on the resume.
Two VP HR candidates. Same title, same Big Ten MBA, same Fortune 500 manufacturing background. Both score 95 percent against the requisition.
One is cut in round two. The other gets the offer.
Go back to the resumes afterward and there is nothing in either one that explains the difference. The screening tool was not wrong. It measured what it was built to measure and it measured it accurately. The problem is that the thing separating those two people was never written down anywhere a machine could read it.
What automated screening is genuinely good at
Worth saying clearly, because the case against over-reliance is weaker if it pretends the tools are useless.
Where it earns its keep
Sorting a high volume of applications into a workable pile.
Confirming hard requirements: certifications, licences, work authorization.
Catching gaps and inconsistencies a human skims past at 4pm on a Friday.
Applying the same filter to every applicant, every time.
Where it fails silently
Any question about how a person behaves rather than what they have done.
Candidates whose resumes undersell them, which describes most strong operators.
Distinguishing between two people who look identical on paper.
Telling you that it missed something.
That last one is the real risk. A screening tool returns a ranked list with no indication of what it could not evaluate, so the output looks complete when it is actually partial. The confidence is the problem, not the technology.
The tool did not fail. It answered the question it was asked. The question was just the wrong one.
The four behaviors that decide it
How they handle being wrong
Ask about a decision that failed. The answer either arrives with a name, a lesson, and an owner attached, or it arrives as a list of external circumstances. Both versions sound reasonable in the room. Only one of them predicts what happens when this person is wrong at your company.
How they treat someone junior with nothing at stake
Every candidate is impressive to the people who can hire them. The revealing question is what happens with the coordinator scheduling the interview, or the analyst who made a mistake at 6pm. Nobody puts that on a resume, and everybody who has worked alongside the candidate knows the answer.
What happened to the people they hired
This is the single most predictive question available in executive HR, and almost nobody asks it. Did their last three hires get promoted, or did they turn over? A leader whose people advance is a leader who selects and develops well. A leader whose people leave has a pattern that will repeat inside your company.
Whether operations trusted them
An HR executive with no credibility on the floor or in the plant will spend three years earning it, and the first eighteen months of that will look like resistance to change. Ask a former operations peer, not a former boss.
How to structure judgment without automating it
The answer to unreliable automated scoring is not unstructured human impression, which is worse. Without a framework, the most charismatic candidate wins, and the gap between interview performance and leadership performance is wider in HR than in any other function, because reading a room is both part of the job and part of the interview.
What works is a defined framework applied the same way to every candidate, in writing, before anyone compares notes.
A structured alternative
Define the outcomes first
Three to five measurable results for the first twelve to eighteen months. Outcomes define the requirements. A job description describes activity that is true of every candidate.
Before any resume is read.
Score the same dimensions every time
Capability, communication style, stability, leadership potential, and long-term fit. Numbered rather than impressionistic, so two interviewers can disagree about a score instead of arguing about a feeling.
One through five, written before discussion.
Ask peers, not only supervisors
A former CFO or COO who sat at the leadership table alongside the candidate saw judgment under pressure. A former boss saw someone who was easy to manage, which is a different thing.
At least one peer per finalist.
Submit before you compare
Everyone writes scores before anyone speaks. The framework survives the room. Impressions do not.
The part people skip.
None of that is anti-technology. Use the tool to build the pile. Use the framework to choose from it. The mistake is letting the tool do the second job because it produced a ranked list and a ranked list looks like an answer.
Why this matters more in HR than anywhere else
A weak CFO shows up in the numbers. A weak COO shows up in operations. A weak HR leader shows up two years later in turnover nobody can explain, a bench that never developed, and a management team that quietly stopped bringing hard problems to HR.
By the time it is visible, it has been true for a long time. Which means the screening decisions made at the very front of the process carry further in this function than in any other, and they are exactly the decisions most likely to be handed to a keyword sorter.
The two candidates, revisited
The one who got the offer had three former direct reports who had since been promoted. The one who was cut described a failed initiative for six minutes without once using the word “I.”
Neither fact was on a resume. Both were findable in about forty minutes of the right phone calls.
Comparing two candidates who look identical?
Thirty minutes, and you will leave with the four questions that separate them, whether or not we ever do business.
Zach Nold is the founder and Chief Talent Officer of Crucial Hire, an executive search firm in Chicago, Illinois that places HR leadership exclusively, from HR Manager through CHRO, for companies across the United States. 312-612-1981 | zach@crucialhire.com
