Edition #82 – Executive Level Search & AI

Hunt Scanlon and Recruiterflow surveyed 97 executive search firm leaders in Q1 this year. 91% of them use ChatGPT or Claude. 20% have AI embedded across multiple workflows, 58% use it selectively, 20% are still poking at it. Among the firms furthest ahead, 74% have wired AI into their ATS or CRM and 68% have built shared custom agents.

Look at where those tools sit and it is all the front half. Mapping, research, drafting, notes. Nobody has moved AI any closer to the decision at the end.

Leadership IQ tracked more than 20,000 new hires across 312 organisations and found 46% failed inside 18 months. 89% of those failures came down to coachability, temperament and motivation. 11% were technical ability. The entire 89% sits in the part of a search that leaves no data trail.

Talent Tip

Write down the three or four decisions the person owns in year one. Whether the Cape Town operation stays open. Whether the ERP goes in before or after the acquisition closes. Whether the ops director who has been there eleven years keeps their team. Then build the interview around those, and ask each candidate what they would need to know before deciding, in what order, and who they would speak to first. Twenty minutes of that beats a full day of competency questions.

Then do your references off-list. The two names a candidate hands you have been briefed. Go two steps out. Someone who reported to them two roles ago. Someone who sat next to them on an exco and has since left. And ask something specific. “What did people bring to them when they did not want to take it to the CEO?” gets you an answer. “How would you rate their leadership?” gets you a paragraph of nothing.

AI Spotlight

I built Aria, so I am not going to pretend AI did nothing here. On the senior searches we closed recently it did four things well.

It built the market map. Every company in the sector at the right size, the person holding the seat, how long they have held it, who they replaced. That work used to sit with a researcher for a fortnight. It pulled the public record on each name into something readable. Annual reports, conference talks, old interviews. I walked into every conversation knowing what the person had said in public about the thing I was about to ask them. It drafted the position spec off the client’s own language instead of a template. And it took the interview notes, which meant I was looking at the candidate rather than at my laptop.

Attribution is the first wall. A CFO’s last business tripled in value. AI can tell you she was there. It cannot tell you whether she caused it or stood next to the person who did, and that is the whole question. The only route to the answer is four calls to people whose names never appear on the CV.

Motivation is the second. Nothing in the public record tells you whether a sitting group exec will leave a business she is loyal to, what her partner thinks about Johannesburg, or that she has already turned down two approaches this year for a reason nobody has written down.

Both of our recent searches came down to two candidates who were, on paper, close to identical. The gap was in how each one handled being disagreed with by a chairman in a live conversation.

Tech Trend

LHH surveyed more than 2,530 companies at the end of last year and found AI is now the number one executive skill gap, with 49% of leaders naming AI and emerging technology a top priority. It ranked seventh the year before. Boards are being asked to hire for a capability most of them cannot assess, and they get fewer attempts at it. High-turnover leadership teams dropped from 43% to 19%.

The market’s answer so far is to rent instead of buy. Heidrick & Struggles and Business Talent Group put demand for interim leaders up 310% since 2020, with interim CFOs making up around half of all C-suite requests. A board will take someone on a nine month contract rather than commit to a permanent hire it has no way of testing.

Personal Insight

AI compressed the part of the search I was already good at and did nothing for the part that keeps me up at night.

I closed the market map in a morning and felt productive. Then I spent three weeks on the phone, and both searches turned on a single reference call each. One was a former direct report who told me, carefully, that her old boss was outstanding in a crisis and restless when things were calm. The client was heading into two quiet years of integration work. That one sentence ended the candidacy, and no tool I own would have surfaced it.

What I have changed is where the saved time goes. Into referencing, all of it. Six or seven conversations per finalist instead of two, most of them with people the candidate did not nominate. That is where the risk sits, and none of it can be automated, because the information does not exist anywhere until you ask a person for it out loud.

If you are running a top-level search this year and your process ends with a competency interview and two supplied references, AI is the least of your problems. Come and talk to me about the referencing first.

Thanks for reading

Joe – AiR