"I can hit it off with anyone over a two-hour dinner." An operating partner said that to me, describing how he evaluates CEO talent. He meant it as a point of pride. It is also, if you sit with it, a fairly precise description of the problem.
A two-hour dinner tells you whether someone is good company. It tells you whether they can hold a room, tell a clean story about their last company, and answer a pointed question without flinching. Those are real signals, and they are not nothing. But they are signals about charm and narrative control, not about whether this person can run this business through this plan. When the original evaluation is a dinner, the data that would actually change the call shows up later. Sometimes on the way to close. Sometimes well after it, when it is expensive to act on.
Charm survives dinner. Execution gaps don't survive the first hard quarter.
The traits that make someone excellent across the table are largely orthogonal to the traits that predict whether they can execute a value creation plan. Confident storytelling, quick rapport, a sharp answer to "what would you do differently" all correlate weakly with what happens when the plan meets a real organization: a finance team that doesn't trust the forecast, a sales leader who has quietly stopped believing in the growth target, a COO who has been managing around the CEO for a year rather than telling him what's broken.
None of that is visible over dinner, because dinner is a one-on-one interaction and the thing you are actually diligencing is a system. A CEO's individual capability is only half the read. The other half is how that CEO operates inside the specific leadership team they will inherit or build, under the specific mandate the deal thesis requires. That is the same insight that applies to reading a management team as a whole: you are evaluating fit against a plan, not competence in the abstract. It is just as true, maybe more true, when the person in question is the one everyone else in the room is calibrating against.
A charming CEO with a weak relationship to the CFO, or a founder-mode operator who has never actually delegated a P&L, will not reveal that at dinner. They will reveal it in month four, when the board is asking why the numbers don't match the story.
The cost of late-arriving data is the call that changes after close
Every deal team has lived some version of this: the CEO who was the reason to do the deal turns out, six months into the hold, to be the reason the plan is behind. Not because anyone lied. Because the diligence process never generated the data that would have shown the gap. The 360 that should have run pre-close ran, if it ran at all, as a formality after the ink was dry. The signal that the CFO didn't trust the growth number, that two VPs were already looking for other jobs, that the CEO's operating style worked in a scrappy startup and was actively wrong for a platform integration, none of that existed as data anyone could act on before the check cleared.
This is the expensive version of the problem. Once you own the company, replacing a CEO costs months of disruption, a search, and a credibility hit with the rest of the leadership team who watched the sponsor get it wrong. The cheap version of the fix is upstream: generate the data before the call gets made, not after. That is the entire argument for building organizational health diligence into the process rather than treating the CEO conversation as a standalone personality read.
What the best diligence conversation I've watched actually looked like
The best CEO diligence conversation I have watched lasted forty minutes. A consulting advisor working the deal had pre-run a 360 on the CEO and the top eight leaders three weeks before the meeting. He walked into the room having already seen where the leadership team's confidence in the CEO was strong and where it thinned out, which direct reports felt heard and which felt managed, and where the CEO's own self-read diverged from how the team actually experienced him.
Forty minutes was enough, because he wasn't discovering anything in the room. He was confirming, probing, and stress-testing a pattern he already had in hand. He asked the CEO directly about a gap the data had surfaced, something the CEO would never have volunteered and the advisor would never have stumbled into with generic interview questions. The conversation was fast because the hard work, the actual discovery, had already happened. In-room discovery is slow and dependent on what the CEO chooses to reveal. Pre-run assessment is fast and independent of what anyone chooses to reveal, because the data already reflects how the team experiences the person, not how the person narrates themselves.
That is the shift. The firms moving fastest on talent have not gotten better at dinner. They have replaced the dinner with infrastructure that generates the read before anyone sits down.
What systematic CEO and leadership diligence actually looks like
The mechanics are not exotic. A 360 on the CEO and the top team, run before the diligence conversation rather than after close, benchmarked against a real comparison set rather than a generic competency model. A read on how the leadership team actually operates as a system: where trust flows, where decisions get stuck, whether the CFO's confidence in the plan matches the CEO's confidence in the plan. Confidence gaps against the specific priorities in the thesis, so "this team feels strong" becomes a number you can defend against 900+ PE-backed portfolio companies rather than an impression from across a table.
Run early enough, this data changes who does the deal, what the first 100 days look like, and what gets written into the CEO's mandate on day one. Run late, it just explains, after the fact, what the dinner missed. Firms that have built this into their process, whether it lives inside the deal team or gets run through consulting partners supporting the diligence, are not doing something more cautious. They are doing something faster: they are compressing weeks of ambiguity about the person who will run the company into a data set that fits inside a forty-minute conversation. For PE firms underwriting a thesis on the strength of a CEO, that is the entire point of diligence: to know before the check clears what the dinner would only have told you after.
The two-hour dinner is not wrong to run. It is wrong to rely on. Charm is real information, and it is the least reliable kind you have. The teams getting this right are not eliminating the dinner. They are making sure it is the last conversation in the process, not the first and only one.
This article expands on a LinkedIn post by Jan Jamrich, CEO of Entromy.