On 6 October 2008, three weeks after the collapse of Lehman Brothers, Richard Fuld told the US Congress that no one had grasped the scale of the crisis, or how the deterioration of mortgage-backed assets would contaminate the rest of the financial system (U.S. House of Representatives).
Yet two years later, the report by examiner Anton Valukas reconstructed risk, liquidity and valuation signals already present inside Lehman: all of it long, long before the financial collapse. He concluded that, in order to maintain market confidence, the company had presented certain aspects of its financial condition in a misleading way (Valukas).
Will we ever know what Fuld really knew? No. And neither the Lehman Brothers documents nor the Valukas report will ever truly tell us more. They are enough, however, to outline a wider problem: namely, that at the top of an organisation reality never arrives in its raw state.
It arrives filtered and interpreted. Often it already comes with an explanation (or a justification).
Power alters the information environment of those who hold it.
That is a fact.
For a CEO, then, acknowledging it is essential: it allows them to look more critically at the information they receive, and to structure channels through which information flows as it should.
The problem is not lies
As early as 1993, Canice Prendergast showed that when performance is assessed subjectively, subordinates may have a rational incentive to conform to their superior’s opinions. The pressure towards consensus originates within the evaluation system, even without explicit orders (Prendergast).
Henry Mintzberg described the manager as a central node in the organisation’s information system. In hierarchical structures, each level selects what moves up. This selection is necessary, because no CEO can receive everything. The risk arises when the criterion of relevance becomes confused with the criterion of approval.
Team members quickly learn which problems the leader listens to, which they play down, which they reward and which cost the messenger time or status.
From that moment, the system starts filtering itself.
Three ways information changes as it moves up
1. Filtering for relevance
Whoever reports information upwards decides what deserves attention. Weak signals and still-ambiguous problems, however, risk getting stuck, because they do not seem mature enough to bring to people with little time.
2. Positive gloss
Many organisations teach people never to bring a problem without a solution. The message is often implicit, but sometimes explicit. A useful rule, certainly: at least until it has the side effect of making every problem look already under control.
So what reaches the top is a critical issue wrapped in reassurances, plans and probabilities. The substance may be urgent; the form makes it look manageable.
3. Converging on expectations
When a leader holds a strong thesis, the data are read within that thesis. Distortion can arise from the selection of metrics, from the way cases are aggregated or from the choice of which scenario to present first.
The person preparing the report can be perfectly honest and still build a picture skewed towards what they think the top will consider plausible.
The simplest indicator: bad-news latency
One of the most useful measures of an organisation’s information quality is the time that passes between when a problem becomes visible at operational level and when it reaches those who can change the decision.
Bad-news latency = date on which the top was informed − date on which the signal was already observable
The formula does not claim scientific precision. It does, however, force a concrete question after every surprise: how long could someone have noticed it?
| Latency | Possible reading |
|---|---|
| Low | The system surfaces signals early, even when they are incomplete |
| Medium | Information moves up after consolidation and verification |
| High | The problem arrives once it has already had consequences |
| Recurring and high | The filter is probably structural, not episodic |
A CEO who often discovers problems only once they become crises does not just have a communication problem. They have an information architecture problem.
The leader’s reaction is part of the system
The question “why didn’t you tell me sooner?” always comes too late if the organisation remembers that the last person to bring bad news was interrupted, blamed or forced to defend it before the data were complete.
That is why the most important question for a CEO is: when someone last brought me information that contradicted one of my beliefs, what did I do in the first five minutes?
That reaction defines the real cost of dissent far more than any statement about a “feedback culture”.
How to build channels that reduce distortion
Cultural solutions help, but they must become structure.
| Mechanism | Function |
|---|---|
| Periodic access to lower levels | Calibrates what middle management filters out |
| “Bad news first” session | Forces problems to be made visible while still incomplete |
| Raw metrics alongside summaries | Reduces dependence on the report’s framing |
| Independent advisor or board | Brings in readings not tied to the internal hierarchy |
| Post-mortems on surprises | Measures when the signal was already available |
| Protected escalation channels | Reduces the personal cost of raising a problem |
The direct channel must not turn into micromanagement. It serves as a calibration tool: the top periodically compares the aggregated narrative with what is perceived closer to execution.
A Business Advisor can be valuable precisely when they remain independent enough to ask the questions the internal chain has learned to stop asking. The condition is real access to data and people, not only to materials already filtered for the top.
The CEO needs worse information, sooner
A healthy information system does not promise that every problem will arrive already verified, quantified and accompanied by a solution. It accepts that some signals will arrive while they are still uncomfortable and incomplete.
The top must therefore distinguish between two errors: being alarmed too early by a false positive and being informed too late of a real risk. Many organisations optimise almost exclusively to avoid the first. That is how they increase the latency of the second.
Power inevitably creates distance. The quality of governance shows in how deliberately the organisation builds ways to reduce it.
Domande frequenti
How do you tell healthy filtering from distortion?
Healthy filtering reduces volume while preserving meaning. Distortion alters meaning through omissions, delays or framing. A key warning sign is the late discovery of problems already known at operational level.
Does the problem worsen with the size of the organisation?
More layers increase the potential points of filtering, but large organisations can compensate with audits, escalation channels, independent data and formal governance. The decisive variable is the quality of the mechanisms, not just size.
How do you measure the latency of bad news?
For every significant surprise, reconstruct when the first signals were already observable and when they reached the level able to decide. The trend over time matters more than any single case.
Does an external advisor really receive less distorted information?
Only if they have independent access to data, people and processes. If they see only the materials prepared for the board, they inherit most of the same filters.
Fonti e riferimenti
- U.S. House of Representatives, Statement of Richard S. Fuld Jr. before the Committee on Oversight and Government Reform, 6 ottobre 2008
- Anton R. Valukas, Lehman Brothers Examiner’s Report: Volume 1
- Canice Prendergast, A Theory of “Yes Men”
- Henry Mintzberg, Managerial Work: Analysis from Observation
- Bliss, Business Advisor: chi è e perché è la figura cruciale che manca alla tua azienda