Doing the Work Is Not the Same as Being Seen Doing It
Most advice about visibility is really advice about volume. Post more. Speak up more. Raise your hand earlier. It treats being unseen as a quantity problem.
The research points somewhere more specific, and more fixable. Your work does not fail to travel because there is too little of it. It fails to travel because, in the places where it lands, nobody can tell exactly which part was yours.
The study that isolates the mechanism
In 2005, Madeline Heilman and Michelle Haynes ran three experiments published in the Journal of Applied Psychology. Participants read about a mixed-sex pair who had worked together on a task and produced a highly successful result, then rated one member of the pair.
When the only information available was about the team’s performance, the woman was rated 5.6 out of 9 on competence. The man was rated 7.3. On influence over the outcome, she scored 5.3 to his 6.5. On leadership, 4.7 to his 6.5. Same work product. Same success. Same written description.
Then the part that matters more. When participants were also given information about individual performance, the gap closed. Competence came in at 8.2 and 8.2. Influence at 7.6 and 7.3. Nobody argued the bias away. The ambiguity was simply removed, and the ratings corrected themselves.
That is one laboratory study, run with undergraduates two decades ago. On its own it would be suggestive rather than settled.
The same pattern, in real careers
The reason to take it seriously is what happened when researchers asked the same question with real careers on the line. Sarsons, Gërxhani, Reuben and Schram, writing in the Journal of Political Economy in 2021, examined tenure outcomes for academic economists. Men were tenured at broadly similar rates whether they published alone or with coauthors. Women were less likely to be tenured the more they coauthored. Two controlled experiments in the same paper reproduced the credit-attribution pattern with the confounds stripped out.
Laboratory and field agree. Shared credit is not split evenly by default. It resolves in a direction, and the direction is not yours.
Now put that next to the shape of the field. The World Economic Forum and LinkedIn’s Gender Parity in the Intelligent Age analysis puts women at 28.2% of the STEM workforce, 24.4% of STEM managers, and 12.2% of the STEM C-suite. Entry is improving faster than the top is. The share of profiles listing AI engineering skills that belong to women rose from 23.5% in 2018 to 29.4% in 2025. More women are doing the work. The narrowing stops well before the level where credit is assigned.
Why this is a Reach problem, not an effort problem
Reach is the distance between doing the work and the work being seen. The usual reading is that you close that distance with volume. The evidence says you close it with resolution.
The question is not how many people saw the output. It is whether the people who saw it could resolve your specific contribution inside it.
That reframes the failure mode entirely. The dangerous condition is not obscurity. It is being present but unresolved: on the team, in the deck, in the credits, with your particular judgment blurred into the group’s. This looks like visibility. It performs like invisibility. And it is the default state of most senior technical work, because most senior technical work is collaborative and most of its outputs are summaries.
Three moves
Assign contribution before the outcome, not after. In the Heilman experiments, the gap closed when team members had worked on clearly distinct tasks rather than overlapping ones. Claiming credit after a success is a fight. Defining ownership at kickoff is an administrative detail nobody contests. On your next project, put in writing who owns which decision, and send it to the group. That one artifact does more for you than a quarter of self-promotion.
Distribute your reasoning, not your results. A result can be absorbed into a team outcome. A judgment call is harder to reassign, because it has a shape only you would have given it. When you share work, lead with the call you made and why you made it over the alternative you rejected. The reasoning is the part that stays attached to your name.
Build a checkable record before you need one. Heilman’s third experiment found that clear, positive information about a woman’s prior performance closed the gap even when the only feedback available was about the group. Prior evidence changes how ambiguous evidence gets read. One owned channel, run consistently, where your thinking is timestamped and attributable, is how that prior evidence is already on file when a room needs it.
You are not competing for attention. You are competing against ambiguity, and ambiguity has a default setting. Removing it is a discipline, and it is a far more tractable one than being louder.
Not sure whether Reach is the dimension actually holding you back? Take the Impact Architecture diagnostic and find out which of the four is your real constraint.
