Cases from the MBA and the Vanguard · MBA module — when a flat organisation sends every decision to the same six people
Subtitle: A flat organisation where every decision climbs to the same six people
A fictional leadership case for discussion.
The managing director had built the branch himself.
When he arrived in Zagreb, the local office of the international group had twenty people and one large customer. Eight years later, it had eighty-five people, a strong regional reputation and some of the best engagement scores in the group.
He was proud of one thing above all. There were no unnecessary layers. Anyone could walk into his office. Anyone could propose an idea. Good people did not have to wait for permission to care.
Then came a Monday management meeting.
The agenda had forty-one items. They ranged from a pricing exception for a key account to the replacement of meeting-room chairs. By item fourteen, the time was gone. Twenty-seven items moved to the following week. Some were moving for the fifth time.
After the meeting, the HR lead put a single page on the table. Twenty-three initiatives had been launched in the past year. Four were completed. Seven were on hold. Twelve were still listed as "in progress," although nobody had touched them in three months. Two senior people had left. One exit interview said simply: "After a while, I stopped proposing things."
He asked the team to walk him through one initiative.
A year earlier, Ana, a senior consultant, had proposed a standard onboarding process for new customers. Everyone agreed it was a good idea. The management table approved it "in principle" and asked her to drive it. Nobody decided when it should be finished, gave her time or asked other teams to contribute. Ana was booked at ninety-five percent on the branch's most important customer.
Twice the initiative came back to the table for small decisions, a template and a three-hundred-euro licence. Both times it was postponed behind more urgent items.
Eight months later, a major new customer started. The kickoff slipped. Three teams asked the customer the same questions. The customer's director complained directly to headquarters. In the review, someone found Ana's half-finished draft. It would have prevented most of it.
"Did we decide it was a priority?" the managing director asked.
"We all agreed it was important."
"That is not the same thing."
Over the following days, he sat with his teams. He found real commitment: people who solved customer problems at midnight and helped each other without being asked. He also found a pattern.
Nobody knew what they were allowed to decide alone. When a senior colleague decided something, she was sometimes overruled a week later, so she began bringing everything upstairs. Pricing exceptions, hiring requests and escalations worked differently each time, depending on who was involved. Missed internal deadlines were never discussed. "People are busy with customers," managers said. "We don't want to blame anyone."
Customer work always won, because customer work was measured. Headquarters tracked utilisation, revenue and satisfaction. Nobody tracked whether anyone had improved their area of ownership. The utilisation target was ninety percent. Ownership was supposed to live in the remaining ten, which did not exist.
Then came the uncomfortable part. The open door led to his office. He liked being involved. When an initiative stalled, he sometimes finished it himself late in the evening, and without meaning to, he taught everyone that stalled work would be rescued. He had never told headquarters that the target left no room for improvement.
"I built a company where anyone can start something," he told his team, "and only I can finish it."
That evening, at dinner with classmates from his COTRUGLI generation, the conversation became more honest. One had written a one-page decision map with clear thresholds, and half her management agenda disappeared within a month. Another had appointed team leads, and a year later had a hierarchy with more meetings than before. A third asked: "When did you last stop an initiative on purpose?" He could not remember.
By Friday, three proposals were on his desk.
The first: introduce team leads with authority over people, priorities and budget within limits. It would relieve the table quickly. It would also cost money, need headquarters' approval and could look like abandoning the culture the branch was known for.
The second: stay flat, but make it explicit. There would be a decision-rights map, one named owner per initiative with protected time, a deadline and a stop date. Only decisions above set thresholds would reach the table, and a quarterly review would formally continue or stop every initiative.
The third: radical focus. All open initiatives would be frozen and three chosen. He would negotiate a lower utilisation target with headquarters for their owners and accept a temporary dip in revenue.
Headquarters expected the numbers to hold. The key account managers would not give up people. The management team wanted its evenings back. Employees wanted to know whether ownership would mean real authority or just more work.
He could combine elements. He could not give everything equal priority. And he still had to decide how much control he was genuinely willing to give away.
Monday's meeting was four days away.
I would open with a show of hands: "Who here has a management meeting where items move to next week for the fifth time?" Most hands will go up. That makes it the room's problem, not the case's.
I would ask the executive from the most hierarchical organisation at the table first. They will likely argue for team leads. That gives the founders and flat-organisation leaders something concrete to push back against, and the debate starts on structure versus explicit rules.
The fact that should change the room's answer is this: headquarters has refused to lower the utilisation target. Once that is on the table, the third option is gone, and the second option has to find time for ownership inside the ninety percent. Watch whether the room still believes ownership is possible, or whether it quietly concludes that the real decision sits with headquarters, and what that says about the branch manager's own authority.
COTRUGLI describes the world we have entered as NEO: networked, exponential, orchestrated. In that world, a branch that routes every decision through six people is not just slow. It becomes a bottleneck at exponential speed. Customers, partners and increasingly AI agents will expect answers within hours, and the number of decisions will multiply faster than any management table can meet.
The NEO turn is that decision rights must become explicit enough to be orchestrated: written, thresholded, and clear enough that a colleague, or a system, can act on them without asking. AI can triage the agenda, track initiatives and flag stalled ownership. It cannot decide who is accountable. The branch that makes its decision logic visible can scale its judgement. The branch that keeps it implicit will only scale its queue.
A flat organisation without clear decisions is not free — it is just waiting for the same six people.
A live case: every round can be improved, and the author's feedback is the next one.
The same two questions, answered twice: first without the mentor's corpus, then from it — the second volume's Command Layer and its failure modes, his decision catalog, VIS and the agent charter.
Without the mentor's corpus
1. The second, run with the third's discipline, and the first only where the map later shows a gap. Jo Freeman saw this branch's trap in 1972, in "The Tyranny of Structurelessness": any group that works together for any length of time will structure itself in some fashion, and where the structure is not written down, it becomes informal and forms the basis for elites. Here the informal structure is six people and one open door. So write the structure down; do not add layers. Then cut the queue. Little's Law says that the time work spends in a system is the work in progress divided by the rate at which it finishes. Nineteen open initiatives finishing at four a year means almost five years for the average one. So on Monday, stop all but three, each with a named owner, a date and a stop date, and in the first ninety days take off the table:
The table keeps what sits above the thresholds, the three priorities and a quarterly decision to continue or stop each one. Ask yourself: which of my decisions are made by a rule, and which only by whoever is in the room?
2. Amy Edmondson draws the map in The Fearless Organization: psychological safety on one axis, performance standards on the other. High safety with low standards is the comfort zone, and that is where this branch lives. People solve problems at midnight and help without being asked, yet missed internal deadlines are never discussed. The way out is to raise the standard and keep the safety, which she calls the learning zone. So put the consequences on the work, not on the person. Every initiative ends in one of three recorded decisions: finished, continued with a new date, or stopped. A stop is a decision; only a silent drift is a failure. Then find the time. Ownership cannot live in a ten percent that does not exist, but the ask to headquarters is smaller than it looks: three owners at a fifth of their time for ninety days is 0.6 of one full-time person, 0.7 percent of an eighty-five-person branch. Ask for that, not for a lower target, and bring the kickoff that slipped, which headquarters has already heard about. And the managing director pays first: the evenings he spent finishing initiatives now go to covering part of the owners' customer work. Ask yourself: what would my people say I do when an initiative stalls?
Sources: J. Freeman, "The Tyranny of Structurelessness", The Second Wave, 1972; J. D. C. Little, "A Proof for the Queuing Formula: L = λW", Operations Research 9(3), 1961; A. C. Edmondson, The Fearless Organization, 2018.
From the mentor's corpus
1. The mentor's second volume lists the ways mission command fails, and two of them describe this branch. The first is intent without boundary conditions: the leader gives intent but not the boundaries that let people read what lies inside their permission. His correction is explicit boundaries, declared "tighter than the leader would naturally want". Nobody here knew what they could decide alone. The second is over-centralisation under pressure: the leader pulls a decision back to the centre, and people who have been pre-empted "stop exercising disciplined initiative even when the pressure subsides". The senior colleague overruled a week later now brings everything upstairs. So the move is the second option, and its instrument is the mentor's decision catalog: every recurring decision gets an owner, inputs, guardrails, an escalation path and a review rhythm. Start with the three that worked differently each time: pricing exceptions, hiring requests and escalations. The same framework asks for authority based on competence, not rank; team leads would add rank before the branch has rules. And the freeze needs what his VIS method calls a kill indicator: an observable threshold, linked in advance to an action. An initiative without one cannot die, and the managing director could not remember the last one stopped on purpose. Ask yourself: which of my decisions have a written boundary, and which only have me?
2. The mentor's Command Layer chapter says how trust is built: it "cannot be declared; it has to be built, across quiet periods, through the deliberate use of shared context, after-action review, and consistent backing of subordinates who have acted within intent and taken losses doing so." So the first consequence falls on the managing director: a decision an owner makes inside the catalog stands, even when he would have decided differently. The second is the review: each quarter, every initiative gets an after-action review and a recorded decision. The chapter also warns that intent does not survive incoherence at the top. People who read intent that says one thing while the leader's behaviour says another "will resolve the incoherence in favour of behaviour". His open door said that anyone could finish; his evenings said that he would. So he stops rescuing, and says so. Then headquarters. In the mentor's doctrine the subordinate answers for the action and the commander for the intent, the boundaries and the authorisation. Time is part of the authorisation, and the managing director never told headquarters that the target left no room for it. That conversation is his, not the owners'. Ask yourself: what does my behaviour teach that my words deny?
On the NEO Turn. The case asks for decision rights explicit enough to be orchestrated. The catalog is exactly that, and it is also the fence for an agent. Under the mentor's agent charter such an agent works at the advisory level:
It never decides. Every routed decision leaves a record of who decided and under which threshold, so the next dispute about who was allowed to decide is settled from the record, not from memory. A decision map an agent can read is one a new colleague can read too.
Sources: Vanguard Leadership, vol. 2, the Command Layer (how trust is built; incoherence at the top; accountability under delegation) and the failure modes of mission command in the full manuscript (intent without boundary conditions; over-centralisation under pressure); the mentor's Vanguard Leadership framework paper (the decision catalog, the review rhythm, competence-based authority); VIS in the mentor's doctrine for intelligence officers (kill indicators); the Vanguard Task-to-Agent Mapping Protocol (the agent charter).
The branch's pride and its problem are the same sentence: no unnecessary layers. W. L. Gore has lived by that sentence since Bill and Vieve Gore founded the company in 1958, and it grew into a global company with no chains of command. Its own presentation of the lattice lists the challenges before the principles: well-defined objectives are needed, decisions must be made, people must know each other, and the compensation process must be fair. Two of its four principles answer this case.
Gore adds a practice: every associate has a sponsor who guides them in growing their contribution. Here each of the three initiatives should have a sponsor other than the managing director, so that a stalled initiative goes to its sponsor, not to his evenings.
Bill Gore put the whole case in one line: freedom requires orderly restraint. The branch has the freedom; Monday's meeting can write the restraint.
Ask yourself: what in my company is really below the waterline, and what have I kept there only out of habit?
Sources: W. L. Gore & Associates, "The Lattice Organization" (the company's presentation of Bill Gore's principles, with its presenter notes); W. L. Gore & Associates, "Our Culture", gore.com.
A question for the table, a disagreement, what you would have done. The case lead reads every comment; the ones the table takes up enter the chapter as questions from the room, with your name.