93 women promoted to manager for every 100 men. Soundssurvivable, almost insignificant. A 7-point gap, dismissed in a single sentenceat a leadership offsite before everyone moves on to the next slide.

‍

So, I ran the math forward, and the math ain’t mathin!

‍

If a 7-point gap at the first promotion were the wholestory, and every rung after that promoted men and women at genuinely equalrates, the damage should stop there. The female share of the pipeline wouldtake one hit, then hold steady all the way up. Starting from parity at entrylevel and carrying a flat 93% relative promotion rate forward, the female sharewould land somewhere around 47% after that first rung and stay close to it.

‍

McKinsey and LeanIn's 2025 Women in the Workplace reporttracks actual representation at every level, not just the first one.

‍

Entrylevel, 49% women.

Manager,42%.

Seniormanager or director, 39%.

Vicepresident, 35%. Senior vice president, 32%.

C-suite,29%.

‍

It does not hold at 47%. It does not hold anywhere. Everysingle rung loses more ground.

‍

Add it up and the number stops sounding survivable. 49% atentry level, 29% at the top. A 20-point drop. A 41% relative loss. Four in tenof the women who should be sitting in that C-suite, by simple proportionalmath, are not there, and no single decision explains it, because no singledecision was ever the actual problem.

‍

Here is where the number gets stranger. Look at the size ofthe loss at each individual rung. (calculated from McKinsey's numbers)

‍

49% (entry) to 42% (manager) = -7

42% (manager) to 39% (seniormanager/director) = -3

39% (director) to 35% (VP) = -4

35% (VP) to 32% (SVP) = -3

32% (SVP) to 29% (C-suite) = -3

If this were simple, repeated bias, the same mistake madesix separate times, the losses should shrink as the pool narrows to a smaller,more scrutinized, more genuinely comparable group of candidates. Screeningtightens at the top. Bias should have less room to operate. Instead, the losseshold steady across every remaining rung, never once tapering toward zero, evenin rooms where the people being evaluated are, by every formal measure, amongthe most vetted talent in the company.

‍

That is not repetition. It is reinforcement, and it has aname in the research. Robert Merton first described it in 1968, and ThomasDiPrete and Gregory Eirich at Columbia formalized it in a 2006 review in theAnnual Review of Sociology. Cumulative advantage occurs when a relativeposition at one point in time becomes a resource that produces further relativegains, rather than a fixed head start that simply carries forward unchanged.The advantage does not just persist through the pipeline. It compounds, becauseholding it changes who gets to make the next call.

‍

Here is what that looks like inside the staircase.McKinsey's own data shows white men climbing from 32% of entry level to wellover half of the C-suite. That is not only the outcome of the broken ladder. Itis also the input to the next rung of it, because the people sitting onpromotion committees at director, VP, and SVP level are drawn from the pool onerung below, a pool that has already been shaped by every rung beneath it. Eachgate is staffed, more and more as you climb, by people who were selected throughthe very gates that produced the unevenness in the first place.

‍

This is not a theory about what evaluators might be doingunconsciously. Lauren Rivera's 2012 study of hiring at elite professionalservice firms found that evaluators ranked cultural similarity to themselves,shared leisure pursuits, background, self-presentation style, and abovetechnical qualification more than half the time. They were not comparingcandidates against a fixed bar. They were comparing candidates against themselvesand calling the result merit. The training that decides what a manager noticesand rewards, the same training Shelley Correll's research traces back towhoever held the pen when the evaluation criteria were written, does not resetto neutral at each level. It gets inherited from the room one floor down, bypeople who were taught to see merit as whatever they already were.

‍

Someone in that room will say the process is not biased, itis meritocratic. Emilio Castilla and Stephen Benard evaluated that claimdirectly in 2010, across three separate studies. When an organization wasexplicitly presented as meritocratic, managers awarded a larger reward to amale employee over an equally performing female employee, not a smaller one.Naming the system meritocratic did not reduce the bias. It gave the bias cover.

‍

That is the mechanism behind Compounded Broken LadderSyndrome, and it is worth naming precisely, because the fix depends on gettingthe diagnosis right. A flat staircase, where every rung loses the same fixedamount, would mean the same bias is simply being reapplied six times, andpatching the worst single rung would flatten the whole curve. That is not whatthe data shows. Each rung is not equally broken. Each rung is more broken thanthe honest description of 7% would suggest, because the people judging it werethemselves selected by the rung below, trained to call their own reflectionmerit, and protected by a label that tells everyone watching there is nothingleft to question.

‍

7 sounds survivable read alone. But that number is not alone.It is the smallest number in a staircase that gets steeper with every step, runby evaluators who were promoted through the same uneven system they now sit injudgment of. Until that mechanism is named for what it is, no single rungrepair will ever hold

‍

‍