The gap in who becomes CEO may be partly a measurement problem. In this study, divisions narrowed it.

You are three months from a succession decision, and the board has asked you for a slate. You open the list of senior leaders and try to write one sentence about what each of them, personally, produced last year. For some names the sentence comes easily. For others you notice you are repeating a general impression, because their results are bundled into a function, a program or a shared outcome that no one can separate out.

A recent study suggests that difference in how legible a person's results are may matter more than most leadership teams assume. The paper is "Firm divisionalization and the gender gap in CEO promotion" in the Strategic Management Journal, by Ulya Tsolmon of the Isenberg School of Management at the University of Massachusetts Amherst and Tingyu Du of the Orfalea College of Business at California Polytechnic State University. The journal's metadata gives a publication date of September 16, 2026. I came to it through a University of Massachusetts Amherst release carried on Phys.org on October 13, so the hook is the coverage; the paper is a month older than that.

How far I got: the paywall kept me from the full paper. Everything below about the method and findings comes from the published abstract and the Phys.org write-up of the university release, and I will say which is which.

The method, per the abstract: longitudinal data on over 616,000 managers in 49,135 US firms. It is an analysis of existing data, not an experiment. The authors argue that divisionalized firms, where units carry their own profit-and-loss accountability, generate more individually attributable performance information about senior managers. They report three results. Women are more likely to be promoted to CEO in divisionalized firms. The gender gap is smaller among managers with prior P&L-accountable experience. And among managers in divisionalized firms, stronger unit performance relative to within-firm peers is more positively associated with CEO promotion for women than for men. The figure that circulated in the coverage — a 37 percent higher probability of internal promotion for women in multi-unit firms than in stand-alone firms — comes from the Phys.org write-up, not from the abstract I read.

This leaves less room for subjective judgments through which gender bias often enters promotion decisions.

Managerial Summary, published abstract, Strategic Management Journal

Here is why this lands differently for the person at the top than for anyone else in the building. Everyone below you is evaluated by someone. You are the person who decides what information the evaluators receive. Structure, reporting lines and which results get a name attached are choices made at the top, and they set the evidence a board or a CEO will later be able to cite. A leader who inherits a structure where contribution is blurred does not see the blur as a bias risk; it simply looks like how the organization works. The study's argument is that the design itself decides whose work can be seen.

The part the research doesn't say, but that I'll say plainly: these are my cautions and my inference, not the authors'. First, this is observational. Firms choose to divisionalize for reasons that may also relate to who gets promoted, and nothing I read shows that restructuring a company would produce more women CEOs. Second, divisions do not fix who is handed a division to run in the first place; if the assignments themselves are uneven, clearer scorekeeping inherits that. Third, I read only the abstract and coverage, so I cannot tell you which controls or robustness checks the full paper reports.

My inference is narrower and does not require reorganizing anything. If subjective judgment fills the space where individual results are hard to see, then any senior leader whose work is bundled into someone else's is being assessed on impression. That applies to any leader you are weighing, not only to one group.

One thing to do this week: take your current list of possible successors, or of anyone you are considering for a larger role, and for each name write down the two or three results you can attribute to that person alone, with the source of each. Mark every name where you could not. Those are the people your evidence is thinnest on, and the gap is yours to close, either by changing how their results are recorded or by saying out loud that you are deciding on less than you thought.

The Leadership Profile starts from the same premise: before you judge how a leader performs, you need a clear read on how they are actually wired to lead. You can see how it works at /wiring-map.