Dan Kalish, Partner at Maven Partnership, shares what he learned running four workshops with executive search leaders on AI adoption.

The same conversation, every time

Last week, I ran four back-to-back workshops with executive search leaders on AI adoption. Different firms, different sizes, different levels of AI maturity. Same conversation, every time.

‍

Someone would describe a tool rollout that had stalled. The room would nod. Someone else would describe a vendor that promised transformational results and delivered a one-hour Zoom training. More nodding. By the third session, I had stopped being surprised.

The pattern is almost universal, and it has almost nothing to do with the technology.

The $200K case study

To anchor the conversation, I used a case study:
A fictional 70-person retained search firm that spent $200,000 on an AI sourcing platform. The CEO saw the vendor at a conference, was impressed, signed a two-year contract, and scheduled a firm-wide onboarding session for the following Monday.

Eighteen months later:

18% platform utilisation. Zero usage among senior partners, consultants, or leadership. The vendor had tried to flag the issue three times. Nobody had returned the calls.

‍

When I put this scenario in front of the room and asked people to diagnose what went wrong, something interesting happened. Almost everyone reached for the same answer: the rollout was too fast, there was no champion, there was no pilot. Correct, all of it.

‍

But the deeper diagnosis took longer to surface, and it was this: the tool was purchased as a substitute for a strategy. Nobody had asked why. Not once, in any structured way, before the contract was signed.

The five whys nobody runs

In each session, we ran a simple exercise. Take a tool you have recently deployed or are considering, and ask "why" until you reach a business outcome. Not a feature. Not a use case. A real business outcome.

It sounds obvious. In practice, it stops most people within two or three questions.

"We rolled out an AI note-taker."
Why?
"To capture calls more accurately."
Why does that matter?
"So consultants can focus on the conversation."
Why does that matter?
Pause. "Because... better conversations lead to better placements?"
Maybe. But is that what you are actually measuring? Is anyone?

‍

In one session, a participant admitted they had adopted three different AI tools in the past year and could not complete the five whys for any of them. Not because they were careless. Because the pressure to "do AI" had moved faster than the thinking behind it.

Why AI change is different

Most people running search firms have managed technology transitions before. The playbook is familiar: get buy-in, run training, manage the transition, measure adoption. AI breaks that playbook in three ways.

‍

The tools are open-ended. When you deployed a new CRM, the system told people what to do. AI tools hand people a toolkit and ask them to figure out what to build. That is twice the cognitive load, and it is why adoption stalls even among people who are genuinely enthusiastic.

‍

The pace of change makes any fixed implementation plan obsolete. And almost nobody prepares for the productivity dip. The partner who tries the note-taker once, finds it slightly clunky, and concludes it is useless has just fallen off the curve at the bottom. Nobody told them the curve existed.

‍

There is a fourth difference that came up repeatedly, and it matters for search specifically: the tools are now accessible to everyone. In prior technology cycles, governance came built in. The CRM told people what they could and couldn't do while AI doesn't. Anyone can spin up a workflow, connect a tool, download (and build!) an agent. The people responsible for data security within your organization are now racing to govern something that moves faster than any policy they can write. That conversation is still early in most firms. It won't stay early.

The adoption failure nobody talks about

The standard failure mode in AI adoption is a tool that nobody uses. There is a second failure mode that is harder to discuss: a tool that the wrong people use, for the wrong reasons, in ways that minimize the thing that makes search valuable.

In one session, a senior partner described why they had never adopted the note-taker their firm had rolled out. Not because they didn't see the efficiency case. Because they feel the best intelligence comes from conversations where a candidate isn't performing for a recording. That is not a resistance-to-change problem but is instead a legitimate professional judgement about where AI adds value and where it erodes it. The firms navigating this well are the ones that have thought carefully about which roles, which call types, and which moments in the search lifecycle should be AI-assisted, and most importantly, which should not.

A conversation that stayed with me

In one session, a participant described a compliance failure at a client firm. An AI writing tool had been rolled out under executive pressure, without proper governance review. The outputs were submitted to a regulator. The result was a serious data breach.

‍

Nobody in the room thought that would happen to them. But the dynamic that produced it was the same one playing out in every firm in that room: executive enthusiasm for AI outpacing governance and risk management. The difference was consequence, not cause.

‍

We spend a lot of time discussing which tools to buy. We spend almost no time discussing which guardrails must be in place before a tool touches client data, candidate conversations, or outbound communications. That conversation is coming, whether firms initiate it or not.

What actually works

After four sessions and several hours of peer exchange, the patterns distinguishing successful from failed adoption were consistent.

‍

Start with process documentation. ‍

Before selecting any tool, map the workflows you are trying to improve. Firms that had documented their search lifecycle, stage by stage, found AI integration dramatically easier. Tools could be applied to specific steps rather than dropped onto an undefined process.

‍

Pick the quick wins first.

‍The most successful rollouts began with a single, high-frequency, time-consuming task and solved it well. Note-taking was the most common example. Once people got two hours back in their week, they became converts, and the champions who sold the next tool to their colleagues.

‍

Build pilots around the people who need solutions, not the people who are already tech-forward.

One participant made this observation in the first session, and it came back in every one that followed. The early adopters are useful. But the more powerful champions are the senior fee earners who were frustrated by inefficiency, who tried the tool under the right conditions, and who can say with credibility: this solved a real problem I had.

‍

Communicate the dip in advance.

Setting explicit expectations that 'you will be slower before you are faster' is not a management concession. It is the single most effective way to keep people on the curve when friction hits. One participant described how her CEO had done exactly this during a CRM migration: she told the team upfront they would be slower and less productive before things improved, and it transformed how the team experienced the transition.

‍

Design the incentive structure.

One firm had introduced a biannual bonus tied partly to visible innovation, something the whole team could see, that moved the business in ways that might not show up in traditional metrics. If your incentive structure rewards individual billings and nothing else, you are asking people to take personal risk for a collective benefit. That is a hard sell.

‍

And respond to your vendor.

Three outreach attempts with no response is not a supplier management issue. It is a signal that adoption has failed and nobody is paying attention.

The real question

The firms with the most traction were not the ones with the most tools. They were the ones where someone with real organisational authority had made a deliberate decision about what AI was supposed to do for the business - and stayed in the room when adoption got hard.

‍

That is not a technology problem. It is a leadership one. The $200K mistake in my case study was not the contract. It was the assumption that a signed contract was the end of the work, rather than the beginning.

‍

To receive a copy of the case study used in the sessions, or to speak with Dan about AI adoption in executive search, get in touch.