The Kickoff Cliff: in the meeting, out of the loop

You’ve been in this meeting. Everyone’s invited. There’s a deck, a timeline, a round of introductions, and a shared sense that this time the project will run differently.

Then the meeting ends, and the project goes somewhere you can’t see.

The leads go off into their own lanes. A few people form a side group. A decision gets made in a thread you’re not on, and you find out three weeks later, usually in a meeting where everyone else already seems to know. Nobody planned it, and nobody set out to shut you out. But the project you were invited to at kickoff isn’t the one that’s actually running.

I’ve been on both sides of this, and I’d bet you have too. I call it the Kickoff Cliff.

Three names for one problem

It helps to name the parts.

• The Kickoff Cliff is the moment: the meeting where everyone has the same picture, and the drop-off right after it.
• Thread politics is the mechanism: who gets added to which email, who gets forwarded, who gets a “let’s take this offline.”
• The shadow org chart is the result: the real map of who knows what. It’s drawn by cc lines, not job titles, and it rarely matches the one HR keeps.

Here’s what that looks like for one made-up team. Switch between the three stages, change the team size, and drag people around.

Stage 2 is the whole story. At kickoff, everyone heard the plan. By week three, the budget changed in a thread with three people on it, and everyone else is working from a picture that’s out of date.

This isn’t just a feeling. It’s a research field.

Researchers have studied this for more than a decade under a less catchy name: knowledge hiding. The founding paper, by Connelly and colleagues in 2012, defines it as “an intentional attempt by an individual to withhold or conceal knowledge that has been requested by another person.” They found three forms: evasive hiding (a partial answer, or a promise to follow up that never comes), playing dumb, and rationalized hiding (“I’m not allowed to share that”). Distrust predicted all three.

How common is it? The number you’ll see quoted everywhere comes from a 2006 Globe and Mail reader poll that Connelly’s paper cites. It’s a self-selected web poll, not a scientific survey, but look at how people answered:

Source: The Globe and Mail weekly web poll, May 12, 2006 (self-selected readers), cited in Connelly et al., 2012.

Only about one in four said “never.” Twenty years later, the same instinct has a new outlet. In a survey published October 1, 2026, 38% of regular AI users said they deliberately keep their best AI prompts and tricks to themselves. It’s one of several recent surveys that asked some version of that question:

Sources: Use.AI via Allwork.space (2026) · Slack Workforce Index (2024) · Microsoft & LinkedIn Work Trend Index (2024) · KPMG & University of Melbourne (2025)

What travels with it

The strongest evidence is a 2023 meta-analysis by Škerlavaj, Černe and Batistič that pooled 131 studies and 47,348 people. It shows what tends to appear alongside knowledge hiding. These are correlations, not proof of cause, but the pattern is hard to miss.

Source: Škerlavaj, Černe & Batistič (2023), Economic and Business Review 25(2), Table 1.

Four things stand out to me:

• Distrust (+.43) and psychological safety (−.47) sit near the extremes. Hiding shows up where people don’t feel safe, and it makes people feel less safe.
• Creativity (−.30) and task performance (−.23) go down. That’s the cost to the work itself.
• Hiding isn’t simply the opposite of sharing (−.06, not statistically reliable). People who share plenty can still hide the one thing that matters. That’s thread politics in a single number.
• It’s a loop. In a separate study of 240 employees, Černe and colleagues (2014) found that hiding triggers “a reciprocal distrust loop.” Coworkers stop sharing with the hider, and the hider’s own creativity drops.

Why people stay quiet

The other side of hiding is silence: knowing something and not saying it. When Milliken, Morrison and Hewlin interviewed 40 employees in 2003, 85% had felt unable to raise an issue with their boss at least once. It’s a small sample, but the reasons will sound familiar.

Source: Milliken, Morrison & Hewlin (2003), Journal of Management Studies.

Most of the time, nobody’s the villain

Here’s the part I want to be fair about. A lot of thread politics isn’t political at all. It’s someone trying to get the work done.

Too many cooks is real. Fred Brooks did the math in 1975: with n people there are n(n−1)/2 possible one-to-one conversations. Five people have 10. Twelve have 66. Thirty have 435. A manager who keeps a thread small isn’t necessarily hoarding power. They might just be protecting everyone’s afternoon.

Some privacy actually helps. Harvard’s Ethan Bernstein studied a huge mobile-phone factory in China. When curtains went up around four production lines so supervisors couldn’t watch every move, those lines’ output rose by as much as 10–15%. Under constant observation, workers had been hiding their better methods. Given a little privacy, they used them. He called it the transparency paradox.

So the goal isn’t “everyone sees everything.” Here’s the line I’d draw:

• Need-to-know protects people or the work: pay, health, personnel issues, drafts that aren’t ready, a small group that needs room to try things.
• Self-serving withholding protects a position: decisions that affect your work, made where you can’t see them, by someone who decided you didn’t need to know.

The trouble is that from the outside, these look identical. When you’re left off a thread, you can’t tell which kind it was. That’s where the distrust starts.

The view from the top looks different

The people who decide who knows what are the least likely to see a problem. Two large surveys asked leaders and individual contributors the same questions:

Sources: SurveyMonkey (2025), 3,573 US workers · Qualtrics 2024 Employee Experience Trends, ~37,000 employees in 32 countries

Leaders aren’t lying. From where they sit, information does flow, because they’re on every thread. The shadow org chart is invisible from the top.

The real culprit: email is addressed to people, not to work

I’ve worked in healthcare, food service and music production, and now I build software. The pattern is the same everywhere. The kickoff is the last time the whole team is in one place, because after that the project lives in email.

Email doesn’t know what a project is. Every message goes to a list of people chosen one send at a time, by whoever hits reply. Every “+Dana” and every “dropping the group from this one” redraws who’s in the loop. Nobody sees the whole map, and nobody owns it. It isn’t a system of record. It’s a pile of private conversations that happen to be about the same thing.

Before email, a message went to a person too. A switchboard operator at the municipal airport in Washington, D.C., July 1941. Photo: Jack Delano for the Farm Security Administration (Library of Congress, no known restrictions).

And it eats the week:

How to stop the cliff

Get rid of the threads, not the people.

1. Give the work an address. Every project needs one shared place where decisions, files and status live, organized by project, team and role instead of by whoever got cc’d. If a decision isn’t written there, it wasn’t made.
2. Make visibility the default and privacy a decision. Open by default. Private on purpose, with a reason (remember Bernstein’s curtains). “Need-to-know” should be a setting someone chose, not a side effect of an email list.
3. Write down roles at kickoff. Who decides, who gets consulted, who just needs to know. Most too-many-cooks problems are really unclear-roles problems.
4. Move decisions out of threads. When something gets decided in email or a DM, whoever decided it posts it where the project lives. One line is enough.
5. Check your own cc line. Before you hit send, ask who’s missing, and whether you left them off on purpose.

None of this needs a reorg. It just needs the work to live somewhere other than people’s inboxes.

Where’s your team on the cliff?

The quiz takes two minutes. It’s anonymous: we ask your role, field and team size so you can compare yourself with others, and we never collect names, emails or IP addresses. Results update live as more people take it. Role and field comparisons appear once a group has at least five responses.

What did you get? Have you been the one left off the thread, or the one leaving people off? Tell me in the comments. I read every one.

Sources

• Connelly, Zweig, Webster & Trougakos (2012). Knowledge hiding in organizations. Journal of Organizational Behavior, 33(1), 64–88.
• Škerlavaj, Černe & Batistič (2023). Knowledge hiding meta-analysis. Economic and Business Review, 25(2), 79–102.
• Černe, Nerstad, Dysvik & Škerlavaj (2014). What goes around comes around. Academy of Management Journal, 57(1), 172–192.
• Milliken, Morrison & Hewlin (2003). An exploratory study of employee silence. Journal of Management Studies, 40(6).
• Bernstein (2012). The transparency paradox. Administrative Science Quarterly, 57(2).
• Brooks (1975). The Mythical Man-Month, chapter 2.
• Microsoft Work Trend Index 2023 and 2024.
• Panopto / YouGov Workplace Knowledge and Productivity Report (2018).
• SurveyMonkey workplace research (2025) and Qualtrics 2024 Employee Experience Trends.
• Use.AI survey via Allwork.space (Oct 2026), Slack Workforce Index (Fall 2024), KPMG & University of Melbourne (2025).