Your Culture Problem Is an Org Chart Problem

Three years ago I sat through a two-day culture workshop. External facilitator, breakout rooms, sticky notes on glass walls. By Friday the leadership team had agreed on four values: collaboration, ownership, transparency, continuous improvement.

Values got printed, framed, and hung in every meeting room. Six months later, the same leaders asked me why nothing had changed. The "collaboration" poster hung in the conference room where I'd just watched three managers argue over which team would get credit for a feature none of them could deliver alone. I asked one question: had anything in the org chart, the bonus structure, or the promotion criteria changed since the workshop?

Nothing had.

The pattern nobody questions

When leaders sense their organization isn't working, "culture" is almost always the diagnosis. The response is always the same: run a program. Hire a consultant. Define values. Launch an internal campaign. Run an engagement survey. Maybe bring in a speaker.

The script doesn't change because the assumption underneath doesn't change: that culture is something you install, like software. That you can debug behavior by updating the messaging.

Here's the problem with that assumption. Language is cheap to change. Structure is politically expensive to change. And culture follows structure, not language.

What actually drives behavior

In every organization I've worked with, behavior is shaped by three structural forces, and they run every day, not once at a workshop.

Incentives. When bonuses are tied to individual performance, people compete. When they're tied to team outcomes, people collaborate. At that company, the annual bonus was calculated individually. The promotion committee evaluated people against their peers. So people attended the cross-team meetings (visible effort) while optimizing for their own team's metrics (actual incentive). The values poster lost to the compensation model every single day.

Decision rights. When a support agent needs three levels of approval to resolve a customer issue, the culture is bureaucratic. When they can resolve it directly, the culture is responsive. Same people. The difference is who holds the authority to act. Organizations that want "ownership culture" while requiring manager-director-VP approval chains have built a structure that makes ownership impossible.

Team boundaries. Component teams (Database Team, Frontend Team, API Team) create handoff culture. Cross-functional feature teams create collaboration culture. The team shape produces the interaction pattern, which people experience as "the way things work here." You can run every team-building exercise in existence, and the Database Team will still hand work to the Frontend Team, because the boundary makes collaboration structurally optional and handoffs structurally required. Reporting lines and career paths reinforce whichever pattern the boundaries set: when promotion means managing a bigger component, managers protect the boundary. When it means delivering customer outcomes, they dissolve it.

These forces operate continuously: every decision, every promotion cycle, every quarterly review. The culture workshop operates once.

Why the fix always fails

Craig Larman, after decades of observing organizational change, formalized what I keep seeing: "Any change initiative will be reduced to redefining or overloading the new terminology to mean basically the same as status quo." The words absorb the pressure for change so the structure doesn't have to.

Every reporting line protects someone's position. Every incentive mechanism reflects a power arrangement. Every team boundary is a manager's scope of authority. Changing any of these threatens someone with something to lose. So instead, the language changes. "Agile" becomes a new name for the same project management process with standups added. "Product Owner" becomes a new title for the same project manager. Values get printed and framed while the bonus formula stays the same.

This isn't conspiracy. Nobody writes a policy saying "preserve middle management at all costs." But hiring decisions, promotion criteria, budget allocation, and meeting structures all independently end up preserving existing positions. The pattern is deniable because nobody planned it. Larman calls it his first law: "Organizations are implicitly optimized to avoid changing the status quo middle- and first-level manager and specialist positions and power structures."

Same workers, different system

Same workers. Same union. Same building. When Toyota reopened GM's closed Fremont plant as NUMMI in 1984, they hired back over 85% of the workforce GM had called its worst: absenteeism between 20% and 45%, cars coming off the line inoperable, thousands of formal complaints.

Toyota changed the system. Workers gained the authority to stop the production line when they detected problems; at GM, that had been punished. Small self-managing teams replaced the foreman system. A no-layoff policy replaced the threat of termination. Over 80 job classifications collapsed into 3. Workers designed their own standard work instead of having industrial engineers dictate it.

Within two years, NUMMI was producing cars at quality levels comparable to Toyota's Japanese factories. Absenteeism dropped to 2-3%. Nobody ran a culture workshop. The system changed what behavior was rewarded, punished, and possible.

But Toyota didn't just swap the structure and walk away. They sent workers to Japan. They taught them why the new system worked, showed them what quality looked like when everyone had authority to stop the line. John Shook, who managed the transition, put it this way: "It's easier to act your way to a new way of thinking than to think your way to a new way of acting." Structure came first. But structure without explanation would have produced confusion, not culture change. The teaching mattered.

What GM learned, and then ignored

GM tried to copy NUMMI's results at other plants. They installed andon cords. They copied the visual management boards. They adopted the Japanese vocabulary.

But they didn't change the system. When workers at other plants pulled the andon cord, they were punished for stopping production. Managers were still paid by volume. At some plants, the cord was literally cut shorter so it was harder to reach.

GM copied the artifacts of the culture. They didn't touch the structure that produced it.

This is the pattern everywhere. Organizations benchmark the visible practices of high-performing companies and import them. Spotify squads. Google OKRs. Toyota andon cords. Then they're puzzled when the behavior doesn't follow. The behavior doesn't follow because the structure didn't change. Nobody wants to touch the structure, because every structural change redistributes power.

The excuse that keeps it going

"Culture eats strategy for breakfast" gives leaders permission to focus on the easy intervention (workshop, poster, offsite) instead of the hard one (changing who reports to whom, what behavior gets promoted, how decisions get made). If culture is the problem, the solution is a culture program. If structure is the problem, the solution is structural redesign. One of those is politically manageable. The other threatens every manager in the room.

The phrase is attributed to Peter Drucker, but the Drucker Institute has confirmed he never said it. What he actually wrote was that "culture, no matter how defined, is singularly persistent." That's a different claim. Culture is persistent because structure is persistent. Structure is the mechanism of cultural persistence. But the misquote strips out the mechanism and leaves leaders with a comforting illusion: that culture is the master variable, and you can work on it without touching the org chart.

The question you're avoiding

If you've sat through a culture workshop, or run one, or are planning one, try this: list the three behaviors you most want to see. Then, for each one, find the structural mechanism that currently rewards the opposite.

"We want cross-team collaboration." Who gets promoted: the person who delivers within their silo, or the person who helps another team ship? If it's the first, the structure is working exactly as designed. The culture is doing what the org chart tells it to.

"We want people to take ownership." How many approvals does it take to ship a decision? If an engineer needs a manager, a director, and a VP to say yes, the structure has removed ownership by design. No amount of encouragement restores what the decision-rights structure has taken away.

"We want transparency." What happened to the last person who surfaced a problem that made leadership look bad? If the answer is anything other than "they were thanked and the problem was fixed," the structure is teaching people the opposite of transparency. And it teaches that lesson every time it happens, long after the workshop is forgotten.

What you can actually do

You probably can't redesign the org chart tomorrow. But you can stop pretending culture is the problem.

Pick one structural contradiction. Just one. The place where what you say you value and what the system actually rewards are most obviously in conflict. Change that. If you want collaboration, tie one meaningful incentive to a cross-team outcome. If you want ownership, remove one approval layer for one type of decision.

I'll be honest about what I'm suggesting. This is a compromise. A real structural intervention would mean changing team boundaries, removing approval layers, redesigning incentive systems. What I'm proposing is narrower: one structural change, focused on one contradiction. But even that is harder than a workshop, because it requires someone with authority to give something up. A bonus formula. An approval right. A team boundary. That's why organizations keep choosing the workshop. Not because it works, but because it doesn't threaten anyone.