Cases from the MBA and the Vanguard · MBA module — when the final veto teaches people to bring problems late (a small pharmaceutical company)
Subtitle: When Authority Becomes the Obstacle to Control
I run the Quality Assurance and Control Department at a small pharmaceutical company. We distribute medicines and medical devices, and we operate a small manufacturing line. Around 120 people work here across office, warehouse, and production floors, and everyone knows everyone by first name. Formally, my authority is absolute and unquestioned: nothing leaves the building without quality release signature. Informally, I have started to realize that this exact authority is our greatest systemic vulnerability.
The Thursday.
Three weeks ago, a batch of an essential medicinal product was finished and ready for final release on a Thursday afternoon. A major hospital tender delivery was due on Monday morning, carrying severe contractual financial penalties for delay and the threat of blacklisting from future public tenders. During final document review, my quality specialist spotted a discrepancy in the batch manufacturing record. It was a missing signature and a mismatched timestamp on an intermediate step—most likely a clerical oversight by a tired operator. But "most likely" is not a parameter on which a regulated pharmaceutical company can sign a certificate of release. Every deviation, regardless of size, must be investigated, risk-assessed, and documented before the product moves.
Properly investigating this single entry required retrieving logs from production, cross-checking physical inventory receipts with warehouse staff, and getting a written statement from the shift supervisor. If everyone responded with urgency, it was a two- to three-day process. Nobody responded with urgency.
Production insisted the entry was a harmless formality and openly complained that Quality was "being rigid and academic again". Sales called me twice within an hour, sent increasingly stressed emails copying executive leadership, and then the Managing Director's executive assistant called to ask if I understood what the hospital penalty would do to our quarterly EBITDA. The warehouse supervisor shrugged and noted they had never been told that specific log mattered at that stage of the workflow. Each department was, by its own internal logic, entirely right. The critical compliance information had simply never traveled between them in a format anyone felt accountable for.
We worked through the weekend, cleared the deviation late Sunday night, and shipped the batch at six o'clock Monday morning. The delivery arrived on time. The hospital was satisfied. Nobody was harmed, and the audit trail was technically immaculate.
The Corridor.
Yet I did not feel we had won. The following Tuesday, a senior sales manager passed me in the hallway and said with a smile: "Well, at least you didn't say no this time." I forced a polite smile back, but when I returned to my desk, the remark stayed with me. I realized with a sudden, uncomfortable clarity that I could not remember the last time a department head had come to Quality before a problem existed. They came to us only when an issue could no longer be swept under the rug—when the clock was running out and the cost of fixing it was at its absolute peak.
They brought problems late because bringing them early meant exposing themselves to a veto.
The Conversation with the Director.
I scheduled a meeting with the Managing Director to address the underlying dynamic. He listened attentively, nodded along, and then offered a response I have been turning over in my mind ever since: "You are doing your job, and nobody blames you for doing it strictly. But I need fewer Thursdays. Fix it however you think best." It was spoken warmly, but it was a complete handover of a structural organizational problem—without a mandate, without a budget, and without an instruction on which trade-offs he was willing to support. I left his office unable to discern whether he was expressing deep trust in my leadership or quietly abandoning me to solve a cultural conflict alone.
The Pattern.
This was not an isolated incident; it was our operating model. Production regularly tweaked equipment parameters during runs and notified Quality after the fact. Sales negotiated delivery commitments with key clients without consulting Quality about release timelines or testing queues. The warehouse learned about quality holds through automated system locks rather than human dialogue. When something went wrong, each department reconstructed the timeline differently. Nobody was lying; each group simply held a different fragment of the truth.
The long-term consequence frightens me far more than the interpersonal tension. In a highly regulated environment, a culture where operational teams view Quality as a hurdle to be avoided is a culture waiting for a catastrophic failure. A company where people plan around compliance is a company where, one day, a critical product defect will be actively hidden rather than reported.
In a regulated business, "we get along" is not a control.
Two Paths Forward.
I see two distinct ways forward, and I believe in the logic of both—yet they point in opposite directions.
The first path is to formalize. Introduce strict mandatory cross-departmental checkpoints before any release-critical date, institute a shared digital deviation log, enforce SLAs for departmental response times, and establish written escalation protocols. This is how a mature, regulated enterprise is supposed to function. It protects the company, it safeguards patients, and it shields my team from political pressure. But I know how it will feel to the rest of the organization: more forms, more gatekeeping, and further proof that Quality is a slow bureaucracy with a stamp. In a 120-person company where personal relationships drive the work, heavy procedure risks replacing conversation instead of supporting it. We might become more compliant on paper, but even less communicative in reality.
The second path is to build trust first. Establish weekly informal operational huddles, embed Quality specialists directly into production planning, and have my team show up on the shop floor before problems arise. I believe this is what genuinely transforms behavior. But it is slow, highly dependent on individual personalities, and leaves the control system fragile. If a key team member goes on leave or a major tender lands during a week when interpersonal trust is strained, we revert instantly to Thursday night. Moreover, I am conscious of a dangerous trap: being "approachable" might subtly train my team to say yes when they ought to say no. Good personal relationships are not a substitute for regulatory controls.
The Pressure.
The pressure arrives from every angle. The Director demands fewer operational bottlenecks. The regulator demands uncompromising documented control. My team wants me to stand firm while simultaneously wanting to stop being viewed as the company villain. Production and Sales want an agile business partner. I am not convinced these expectations can be reconciled under our current structure.
I do not know whether the answer is to make Quality harder to avoid or easier to approach. I have not decided, and I want to hear from leaders who have run organizations where the person holding the final veto was also the person nobody wanted to call.
As organizations automate quality and compliance gating through AI agents and continuous digital auditing, the "veto" moves from a human desk to an automated algorithm. However, automated gates do not eliminate human workarounds—they accelerate them. If operational teams view an AI compliance system as an unyielding wall, they will learn to feed it manipulated input data, delay logging parameters, or bypass digital checkpoints altogether until the system collapses at scale.
How do you design autonomous compliance agents that act as collaborative advisors rather than silent walls? When control moves at machine speed, how do you prevent human teams from treating compliance as an adversary to be outsmarted?
“A control system that people learn to avoid is no longer a control—it is an invitation to a disaster you will only read about in the postmortem.”
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 — Vanguard Leadership, his decision doctrine, the Command Layer, VIS, the teleology magnifier and the agent charter.
Without the mentor's corpus
1. Formalise first, but formalise the early report, not the late gate. The PIC/S guidance on data integrity distinguishes "open" cultures, where hierarchy can be challenged and full reporting of failure is expected, from "closed" ones, where reporting failure is culturally harder. A closed culture, it says, may need more oversight and secondary review for the same level of control. So formal control is not the opposite of trust; it is the price of its absence. James Reason, drawing on the designers of NASA's aviation reporting system, listed what makes people report, among them separating those who collect reports from those who can impose sanctions. Here, the desk that hears the warning also stops the shipment. So make an early report the way to avoid a hold, not to trigger one, and treat lateness itself as a finding. Then expect a paradox. Amy Edmondson found in 1996 that the hospital units with the strongest leadership and relationships recorded more medication errors, because detected errors depend on reporting. Reported deviations should rise before Thursdays fall. The choice breaks down at the signature: approachability changes when Quality hears of a problem, never what it decides. Ask yourself: when someone brings me a problem early, what happens to them first?
2. As spoken, it is closer to abdication, because it hands over what the industry's own standard keeps with him. ICH Q10, the guideline on pharmaceutical quality systems, gives senior management "the ultimate responsibility" for an effective quality system. It asks management to define every unit's authorities and inter-relationships, ensure timely escalation of quality issues, and commit resources. That is the mandate, the trade-offs and the budget he left out. Escalation even ran backwards: the call about EBITDA came from his office. The compliment felt like trust and abandonment at once for a reason: Cheong and colleagues found in 2016 that empowering leadership both builds confidence and adds job-induced tension. Do not refuse the handover; return it on one page he can sign:
That is the mediation he owes: owning the revenue side of every quality hold. Ask yourself: which of my responsibilities came without the authority to make it stick?
Sources: PIC/S PI 041-1, "Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments", 2021, §6.1.3–6.1.4; J. Reason, "Achieving a safe culture: theory and practice", Work & Stress, 1998; A. C. Edmondson, "Learning from Mistakes Is Easier Said Than Done", Journal of Applied Behavioral Science, 1996; ICH Q10, "Pharmaceutical Quality System", 2008, §2.1; M. Cheong et al., "Two faces of empowering leadership: Enabling and burdening", The Leadership Quarterly, 2016.
From the mentor's corpus
1. The mentor's first volume answers this with security as its example of sheepdog work. The wolf approach punishes mistakes and builds fear-based compliance: people hide problems, and incidents are reported late or not at all. That is the Thursday. The sheepdog approach creates the safety to report immediately and keeps protocols clear, easy to follow and regularly updated. The sheepdog does not choose between procedure and approachability; it needs both, because security is a system property, not an individual one. The mentor's decision doctrine names this company's flaw: a system that depends on one permanently trustworthy guardian is badly designed, and multiple imperfect checks are preferable to one supposedly perfect guardian. Move small checks to where the work happens: the record signed before the shift ends, the receipt confirmed the day it arrives, the release queue checked before a date is promised. Keep the rules between departments within the mentor's limit of five, simple enough to explain from memory. The choice breaks down when a procedure fails that test, or when approachability starts to bargain over the release. Ask yourself: how many checks stand between an error and a patient, and how many of them depend on me?
2. It is half of an intent, and the missing half was his to give. The mentor's first volume builds intent in four moves: state the outcome, explain why it matters, name the constraints (what we won't compromise), and leave the how open. "Fewer Thursdays" is the outcome; "fix it however you think best" leaves the how open. The constraints are what was missing: the trade-offs he would back. The Command Layer chapter of his second volume calls intent without a mechanism for distributed execution a slogan. The chapter on training draws the harder line: a leader who issues intent to people who cannot act on it has not decentralised the institution but abandoned it. Quality can act on quality, not on what Sales promises or how Production runs. So complete the intent: take it back with the constraints written in and a kill indicator from the mentor's VIS method, an observable threshold linked in advance to an action. The assumption: Quality can fix this without his authority. The tripwire: a second Thursday in one quarter. The action: he convenes Sales, Production and Quality and decides the trade-offs himself. Ask yourself: what would prove my mandate too small, and who acts when it does?
On the NEO Turn. The mentor's teleology magnifier explains why automated gates accelerate workarounds: AI is never an ethically neutral optimiser; it amplifies the purpose and behavioural pattern that guide its deployment. Automate a veto people already avoid, and the avoidance scales. Change the pattern first, then write it into his Agent Charter:
His first volume adds the record that ends fragmented truth: every meaningful event between parties must produce a shared, co-attested object that neither party can unilaterally alter afterwards. A late or edited log then shows itself.
Sources: Vanguard Leadership, vol. 1, "Security as Sheepdog Function", §4.2 "Creating Conditions for Emergence" and "The missing layer"; the mentor's decision doctrine of 22 August 2026; Vanguard Leadership, vol. 2, the Command Layer and "Training Under Conditions"; VIS in the mentor's doctrine for intelligence officers, chapter 4 (kill indicators); the mentor's glossary, the teleology magnifier; the Vanguard Task-to-Agent Mapping Protocol (the Agent Charter).
The line that matters most in this case is spoken in the corridor, not on the Thursday: problems came late because coming early meant facing a veto. The mentor's case study of Bayer finds the same pathology in a 160-year-old pharmaceutical company, and names it the wolf structure. It rewarded the visible and punished the invisible. A manager who hit the quarterly numbers got promoted; a person who prevented a disaster nobody saw never got credit. People worked hard to avoid being fired, not to solve real problems. The author is the person who prevents disasters nobody sees. Her Thursday ended with an immaculate audit trail and no credit, and the corridor remark shows what the company counted: that she did not say no.
What Bayer changed. Bayer's answer was not fewer controls. Its chief executive announced that small, self-managed teams focused on a customer or a product would take over 95 percent of the decision-making from managers, with progress measured in rapid 90-day cycles. A regulated company can do this because the release decision is not among the 95 percent; it stays where the regulator puts it. What moves is everything around it: the planning, the promised dates, the handovers. Those are the places where the Thursday was made.
How I would run the first ninety days. For a company of 120 people, that means one team and one cycle:
None of this relaxes a single control. What changes is when the truth arrives. Ask yourself: who in my company prevented a disaster this quarter that nobody saw, and how would they know I noticed?
Sources: the mentor's case study "From Control to Connection" (Bayer's Dynamic Shared Ownership); Bayer, Q3 2023 media update (Bill Anderson on self-managed teams, 95 percent of decisions and 90-day cycles); Vanguard Leadership, vol. 1, "The Invisible Wins" (p. 386).
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.