The Work Nobody Owns
When Reliability Becomes a Liability
“Bwip-bwip.”
You know that sound. It’s 4:47 p.m., and with one Teams message, your hope of logging off on time disappears.
“Quick favor, can you clean up this deck for tomorrow?”
When you open the file, you’re met with all the signs of a chop job: 32 slides, half of them broken images, “WIP” across most of the pages, someone else’s speaker notes, three identical slides, and one slide from a completely different project.
You message back. “When do you need this by?”
15 minutes later. “The presentation is tomorrow at 9am. I really appreciate your help.”
Several hours later, while eating dinner at your desk, you finally finish.
You didn’t write the deck. Your name isn’t on it. You won’t even be in the room for the presentation. It wasn’t your responsibility, and you might get a shout-out, but your name isn’t on the deck. The hours you invested aren’t evident to anyone but the person who asked. This surely gets you an “exceeds expectations” in your end-of-year review, right?
It doesn’t
The work in the scenario does not show up on any goal. It does not show up in a capacity model. It happens because someone in the organization got lost in the weeds and didn’t know how to put together their story, while you are using context clues to create a manageable narrative.
The person being tapped to untangle this mess is rarely the most senior. They are the most reliable.
Non-promotable tasks
Babcock, Recalde, Vesterlund, and Weingart published a paper in the “American Economic Review”1 in 2017 with hard numbers. Women volunteer for non-promotable tasks roughly 50% more often than men. They are asked to do them 44% more often. When asked, they say yes about 50% more often. The paper describes these as low-promotability tasks, meaning work that does not advance the career of the person doing it. That gap has not closed as of 2026.
Tanya Reilly, in a 2019 talk that became required reading on engineering teams, calls a related pattern “glue work.” The connective work that holds a team together. Documentation, follow-ups, mentoring, cross-team alignment, project tracking. Reilly’s frame is gentler than the one I am about to give you, because she is talking about engineers, and engineers eventually get promoted on craft.
Program managers do not. Operations people do not. The person on the other end of that 11th-hour Teams message does not.
Glue work is what happens when the connective work is least recognized. Invisible labor is what happens when it is neither recognized nor rewarded.
The failure is structural. The organization has a deliverable due at 9 a.m., and the work gets done because someone absorbs it. The cost of that absorption is paid by one person. The benefit of the deliverable is credited to someone else. This is not a kindness problem. It is an org design problem dressed up as a kindness opportunity.
If the work is not in the JD, it is not in the role.
If it is not in the role, it is not in the capacity model.
If it is not in the capacity model, it is not real to leadership.
If it is not real to leadership, it is not on the performance review.
It is real to you. You did it.
Does this sound familiar?
Maybe you’re the go-to person who gets the last-minute ping.
Maybe you were the one who pulled the deck together while someone else presented.
Maybe you were the program manager doing the “admin-y” work while also moving the teams forward.
Maybe you have also been the person on the other side of that Teams message.
Maybe you sent it because you were underwater, the deadline was real, and the options limited.
Either side of that Teams message, you are inside the same operating model. The model runs on absorption. The absorption is not on any slide.
How to break the cycle.
The next time your Teams pings near quitting time, send this back.
”Sure, I can help. Here are my top three projects: [list of high-priority, high-visibility projects] What am I pushing off?”
Trade-offs are real
You can’t add hours to your week. You also aren’t refusing to help. It is a capacity question. The person on the other end has to engage with it, in writing, in a tool that keeps receipts.
Three things happen when you send it.
The first is that the asker has to look at your plate. Most asks are sent by people who have no insight into your workload, because nobody has built a system that makes your week legible to them. The question forces a brief, polite conversation about priorities, and how “THIS ASK” isn’t one of them.
The second is that the answer, whatever it is, becomes data. If they say “This is the priority, drop something to make room” you have a written instruction to deprioritize. If they say “no, do all of it,” you have written confirmation that the work is being requested on top of capacity. That is a different conversation than “you didn’t deliver.” If they say nothing, the silence is also data. They went to find someone else. 🤞
The third is the slowest and the most important. The first time you send it, it costs you something. The fifth time, it costs you nothing. Eventually, the people who are asking for the support stop asking you, because they have learned that you will surface the cost of the absorption, in writing. They will send it to someone else. That other person has not learned the script yet.
The job is not to refuse the work. The job is to refuse to absorb it.
Indispensable is a trap with better PR.
The Operating Gap diagnoses the systems that bury good work. If this landed, forward it to the person who still hasn’t sent that message. Or send me your story at raevyndigital.com.
Babcock, L., Recalde, M. P., Vesterlund, L., & Weingart, L. “Gender Differences in Accepting and Receiving Requests for Tasks with Low Promotability.” American Economic Review, 107(3), 2017. doi.org/10.1257/aer.20141734