Say a nominating committee is down to two finalists for an open board seat. One spent a decade running security at a company that was breached, and talks about risk in the flat, specific way of someone who was blamed for it in public. The other worked in operations at two companies a lot like this one and gets along with everyone she meets. Over dinner the chair asks the committee how each of them would do in the room. By dessert that question has made the decision, and the conversation has moved on to whether Thursday works for the announcement.

What 81 percent said

PwC's 2026 Annual Corporate Directors Survey came out in mid September and repeats a finding from the year before: 55 percent of directors think someone on their own board should be replaced.1 Among those directors, insufficient expertise is now the reason most often cited, by 39 percent, up from 21 percent in 2025, although PwC also reworded that answer between the two years.1 PwC also asked what matters when evaluating prospective directors. Cultural alignment is very important to 81 percent of respondents. Specialized expertise, such as AI or cybersecurity, is very important to 27 percent.1

A word on the source. PwC audits public companies and sells governance advice to their boards, so it has an interest in boards seeing room to improve. The full report says 599 directors took part.1 These are directors describing their own boards, which tells us what they believe and less about what any board actually does.

Two findings in one report

The survey does not test whether those findings are linked, and I will not claim it does. The question about candidates looks ahead to the next seat a board fills, so it describes today's priorities, and the two figures come from different groups of directors. Fit also cuts both ways in the same report: 27 percent of the directors who want a colleague gone cite an interaction style that hurts the board's dynamics.1 My read is still that the replacement finding and the fit finding belong to a single story. Directors want a colleague gone, and the reason they give is increasingly a missing skill. Directors as a whole say, by better than three to one, that the people they bring in should fit. If the screen for a new seat asks first whether someone will work comfortably with the people already there, a skills gap is what you would expect to find a few years later. AI is the skill directors most often say their boards need to strengthen, at 71 percent.1 A search could have been built to close that gap, had expertise outranked alignment in the criteria.

A fit screen applied to every candidate produces a board that resembles itself in the very habits its own evaluations were supposed to examine.

The assessment numbers lean the same direction. Almost half of directors, 49 percent, say board assessments are not sufficiently candid, and 73 percent say their assessment processes could improve.1 Candor depends on the person speaking and also on the room receiving it, and a room assembled for comfort with its own manner has less candor available. The fit that directors were chosen for includes how an unwelcome observation gets delivered to a colleague of long standing.

Who gets to define fit

The mechanism is old. In a 1995 study in Administrative Science Quarterly, James Westphal and Edward Zajac followed 413 Fortune and Forbes 500 companies from 1986 to 1991. Where the CEO held more power than the board, new directors tended to resemble the CEO demographically. Where the board held more power than the CEO, new directors tended to resemble the existing board. Greater similarity between the CEO and the board went with more generous CEO pay contracts.2 The study is more than thirty years old and measured demographic similarity, a different thing from the cultural alignment PwC asks about, so I offer it as precedent for the mechanism and nothing stronger. Whoever controls selection decides what the room is supposed to resemble, and resemblance is easy to describe as chemistry.

What the executive in front of that board should ask

Anyone presenting an AI plan to a board is presenting to directors who mostly say their board needs to strengthen its AI skills to oversee it well. That puts weight on the presenter. The plan has to be explainable to a smart outsider, with the assumptions written down where someone outside the specialty can challenge them.

Alignment is a legitimate thing to want in a director, since a board that cannot work together governs very little. The trouble starts when it is checked first, in every conversation, with every candidate, and expertise gets examined after the committee already likes someone. A committee could reverse the order. It could write down, before the first interview, the one capability no current director has, and treat comfort as a question for the final round. That is how I would run it, and the survey does not say whether anyone has.

In late September I wrote about the research on how ties among directors hold back dissent once they are seated. That piece covered the people already at the table. Who gets a seat is decided a step earlier, at the dinner.

When your board last filled a seat, what did the committee write down as the thing the new director had to be able to do that no one at the table could?