Most organizations invest in two things to build honest self-assessment: psychological safety and process design. Both are necessary. Neither is sufficient. A third discipline — deliberate, unannounced verification — is what tells leaders whether the first two are actually working, or only appear to be.
The Blind Spot in Two Good Investments
Organizations that invest seriously in trust and in process design still get blindsided — not occasionally, but predictably. The failures aren’t failures of culture or of design. They show up as a green dashboard, a clean quarterly self-assessment, a satisfactory rating returned so consistently that no one remembers the last time it said anything else.
The standard explanation is that the culture or the process wasn’t good enough. I’d argue that’s usually the wrong diagnosis. The real gap is a third discipline that most leadership frameworks treat as redundant once the first two are in place: independent, unannounced verification.
I spent 28,500 people’s worth of career learning this the first way, and a 680,000-person organization’s worth learning it held at scale — as the second-ranking officer in one of the largest workforces in the world. The size changes the stakes. It doesn’t change the lesson.
Trust and Process Are Not the Same Discipline as Verification
The dominant models for building organizational honesty split into two camps. One focuses on interpersonal trust — safety, conflict tolerance, the conditions under which people are willing to say the uncomfortable thing. The other focuses on system design — incentive structures and process discipline built so that honest reporting is the natural output rather than an act of courage.
Both are foundational. Neither answers a separate question: is the reporting system, as it actually operates today, telling the truth? Trust and process describe the conditions under which honesty becomes possible. They don’t confirm that it happened. That confirmation requires a distinct, third discipline — verification — with its own method, not a byproduct of getting the other two right. I don’t grade the report. I grade the room.
A Perfect Record Is a Red Flag, Not a Result
Early in my career, I reviewed a self-assessment process that had returned a clean result for years — every unit, every cycle, rated ready. Not one had ever failed. On its face, that looked like organizational success.
I read it differently. A self-assessment that never once returns an unfavorable result isn’t evidence the standard is being met. It’s evidence the test has stopped functioning as a test. When the only way to fail is to self-report failure, and self-reporting failure carries a cost, the system will approach 100% favorable results regardless of the underlying truth. This isn’t a claim about anyone’s integrity. It’s a structural property of self-graded systems under any incentive pressure at all.
The response was to reinstate unannounced verification — checks with no advance notice and no opportunity to prepare a favorable answer. The very next cycle surfaced real deficiencies. That was the intended outcome, not a failure of the rollout.
Verification, in practice, isn’t complicated, but it has a specific method. Walking into an unannounced review, the signal isn’t in the prepared briefing — it’s in what happens around it. Rehearsed answers have a different rhythm than honest ones. A team that’s confident sounds different from a team that’s been coached. And the most reliable question I’ve found is rarely the one on the agenda; asking something that has nothing to do with the briefing — when someone last took leave, for instance — tends to surface more truth than any question the room prepared for.
The same standard scales down as well as up. I once inherited an operating unit that was failing its own performance standard on every visible measure — and gave the two people accountable for it sixty days to fix it, with no ambiguity about the consequence of missing that date. It was fixed in fifty.
It also generated legitimate resistance at the larger scale. Some of the concern was substantive: people closest to the work read the new standard as a signal of distrust, and that concern deserved a real response rather than dismissal. The resolution was to revise the mechanism — frequency and method — while holding the underlying standard fixed. The standard should not move. How it is verified should be adaptable, and treating the two as inseparable is how organizations end up defending a rigid process rather than a real standard.
Why Verification Has to Be Unannounced
A well-established idea in manufacturing quality management — go to the actual location, observe the actual conditions directly rather than relying on the report — is close enough to this argument that it invites a direct comparison. The distinction is conditional, not conceptual: that method assumes a foundation of trust already exists, such that a scheduled, routine observation is sufficient to see the truth. That is the right approach for organizations that have already earned it.
For organizations that haven’t yet established that foundation — or believe they have and haven’t tested the belief — a scheduled check will be prepared for, consciously or not. If verification requires a calendar entry, that itself is diagnostic: it indicates the organization has not yet reached the state where a scheduled check would be sufficient.
Two Objections Worth Taking Seriously
Two well-known critiques of quality management deserve a direct answer, because both anticipate this argument and both are worth engaging rather than avoiding.
W. Edwards Deming argued that organizations should cease dependence on inspection to achieve quality, and build quality into the process instead — inspecting for defects after the fact is waste, and a sign the process itself has failed. Read narrowly, unannounced verification looks like exactly the inspection regime Deming warned against.
I’d argue it isn’t, for a specific reason: Deming was describing inspection of outputs — checking finished products for defects that should have been designed out. Verification, as I’m describing it, doesn’t inspect outputs. It inspects the integrity of the self-reporting system itself — whether the measurement can be trusted at all. Deming was equally opposed to management by target and by slogan, and equally distrustful of numbers divorced from real process knowledge. Read that way, this argument is closer to Deming’s own skepticism of hollow metrics than to the inspection regime he was arguing against.
Russell Ackoff’s critique cuts differently: a system that rewards the wrong behavior will produce the wrong behavior no matter how well you inspect it, and a good inspection regime bolted onto a bad incentive structure is solving the wrong problem. This one I concede in full, because it’s correct. Verification does not repair a broken incentive structure. What it does is tell you, reliably, whether the system you believe you’ve built is the system that’s actually operating. That’s a narrower claim than “this fixes culture” or “this fixes incentives.” It’s also a more honest one, and it’s the reason verification has to sit alongside system redesign, not substitute for it.
Applying the Standard to Its Author
The standard underlying this argument is one I’ve stated plainly to everyone who has ever worked for me: accountable, not responsible. Not the effort. Not the process. The outcome, regardless of the stated reason for a shortfall.
I have not always applied its inverse as consistently: claiming credit as directly as I have assigned accountability. Redirecting credit upward — to a team, to leadership above you, to the process — is easy to justify as humility, including in instances where the decision and its risk were mine alone. It is worth naming plainly, because a standard that one enforces on others but exempts oneself from is not yet a complete standard. It is a partial one.
Takeaways for Leaders
- Trust-building and process design are necessary conditions for organizational honesty. They are not sufficient, and treating them as sufficient is a common and consequential error.
- A perfect or near-perfect self-assessment record over time should be treated as a signal to investigate the measurement system, not evidence of organizational success.
- Verification should be unannounced until an organization has demonstrated, not assumed, that its culture supports honest routine reporting.
- The standard being enforced and the method of verifying it are separate variables. Hold the first fixed; allow the second to adapt under legitimate challenge.
- Apply the accountability standard to credit as rigorously as to blame. A standard exempted from its author is incomplete.
If nobody told you your numbers were good, would you still believe them? Trust what’s tested. Not what’s told.
Scott Pleus is an operating executive who has built his career on a single conviction: an organization’s confidence in its own readiness is only as good as the last time someone actually tested it. If nobody told you your numbers were good, would you still believe them? That conviction was forged over three decades of command in the United States Air Force, retiring as a Lieutenant General and the longest-serving Acting Vice Chief of Staff in the role’s 79-year history. His command experience is the civilian equivalent of running large, high-consequence operations at every level of scale — from a single division, to a joint international operation, to the number-two position in a 680,000-person global organization. In every one of those roles, the product was readiness, and there was no such thing as a plausible excuse for being wrong about it. His standard, in his own words: accountable, not responsible. That is the experience most executive backgrounds don’t have, and it is what any organization needs when it wants to know — not hope — that it can perform under pressure. Early in his career, he noticed something worth learning from: a self-assessment that had never once produced a failure. Rather than take that as good news, he treated it as a signal that the test itself needed to evolve. He built a new version of the check — one that measured real readiness instead of rehearsed readiness — and kept refining it, even when the organization pushed back, because he was committed to the standard underneath it, not just the version of it that was easiest to pass. That same instinct shaped how he led. He made a habit of showing up unannounced, not to catch anyone, but to see the organization as it actually ran day to day — and to ask the one question a prepared briefing never answers. That habit paid off when he inherited a struggling unit: he gave its two senior leaders a clear deadline and clear ownership of the turnaround, and in sixty days, the unit had rebuilt the standard for itself. The stakes in his operating environment left no room for a wrong answer — there was no version of a readiness failure that was merely inconvenient. That is precisely why the discipline transfers: an executive who has enforced a verification standard with zero margin for error brings a rigor to operating risk that most executive backgrounds never have to develop. Trust what’s tested. Not what’s told.
