The Two-Organizations Problem

A few months ago, I sat in a board meeting at a financial services firm where the CEO was walking the directors through the company’s AI transformation. The slides were impeccable, detailing AI adoption rates by function, time saved per workflow, and a clean line from pilot to scale. The board seemed satisfied that they had the information they needed.

Just 24 hours earlier, I had sat with the engineers who actually built those tools. What they described was a different company. Two of the most senior engineers had already started looking for new roles, citing the gap between what was being claimed in the room I had just left and what they experienced every day.

Both versions of the AI transformation story were true, but neither was complete. This example illustrates what I call the two-organizations problem. I have seen the same pattern at companies across financial services, fintech, enterprise software, and AI, both from the inside during my years as an executive and now from the outside in my advisory work with boards and CEOs. Industry varies. Size varies. The pattern does not.

Every company that has reached a certain level of operational maturity and complexity—typically where there are multiple layers of leadership, established processes, and a need to balance alignment with autonomy across teams—is, in fact, two organizations operating in parallel:

• The reported organization is the version that exists in dashboards, board materials, town halls, and analyst calls. It is the version leaders see, govern, and explain to the world.

• The lived organization is the company as employees experience it day to day, encompassing the quality of the work, the state of the team, the customer experience, and the culture.

The divergence is not a communications failure, but a structural feature of how scaled organizations work. And in some organizations, the gap is widening.

The consequences slowly accumulate: Strategy gets set on a picture that has been shaped more by the reporting process than by the underlying reality. Operational problems surface late, often during a crisis the reported organization could not see coming. The talent closest to the lived organization, the people who feel the gap most acutely, are also the most likely to leave. And boards make succession decisions based on what they have been shown rather than on what is true.

Why the Divergence Is Structural

The two-organizations problem doesn’t stem from individual leaders making bad decisions. Three forces produce this gap.

The first is information layering. Every reporting layer between the frontline workers and the C-suite performs a small act of summarization, smoothing, and selection. Most of these acts are reasonable. A team lead aggregates 10 signals into a single number for her manager. The manager folds that into a function-wide summary. The function head presents a polished view to the executive committee. By the time the picture reaches the CEO, there have been dozens of judgment calls about what mattered and what could be left out. None of those judgment calls were inherently wrong on their own, but their cumulative effect is an organizational view that has been systematically polished.

The second force is incentive shaping. People present what they believe will be rewarded. In organizations that reward green dashboards, dashboards are green. In organizations that punish surprises, surprises stop. In organizations where the senior leadership wants to hear that the AI strategy is working, every report that crosses the executive’s desk notes that the AI strategy is working. This is human behavior responding to signal.

The third is tenure curation. The longer a CEO has been in their role, the more deliberately the information environment has been curated and calibrated around what they are known to find useful, threatening, or unwelcome. In year one, the CEO would have seen one organization, but by year five, the CEO is governing the version of the company that the organization has learned to show them.

AI is accelerating all three of these forces at once. The reported organization can be generated, summarized, and polished faster than ever. Outputs that once required hours of analyst work can be assembled in minutes. The signals that used to leak through—the rough drafts, the off-the-cuff hallway updates, the imperfect first pass at a deck—are being filtered out before any human eye crosses them. The reported organization is becoming more compelling and more difficult to puncture.

A specific version of this dynamic has emerged in the past 18 months across the enterprise software companies I advise. Generative AI tools are being layered into the reporting stack at exactly the moments where data crosses upward. The board deck is now drafted by an AI assistant working from the function head’s prompt. The function head’s prompt is itself an artifact of what they have learned the executive team wants to hear. The presentation that arrives for the CEO is more coherent and more authoritative than any prior version. Whether it is more accurate is a question almost no one is asking.

Where the Gap Is Widest

The two-organizations problem is present in every scaled company, but it is not equally severe everywhere. Four conditions predict where the gap is most consequential:

• The scale of a function relative to the leader’s prior experience. A CEO who came up through go-to-market may have an exceptionally accurate sense of what is happening in revenue and a very shaped sense of what is happening in engineering. The gap is often widest in the functions the CEO has not personally lived in.

• The time since the leader last did the actual work. A CFO who has not closed books in 15 years is governing a finance function that has materially changed in tooling, headcount, regulatory complexity, and pace. The institutional memory is intact. The operational fluency has aged.

• The number of reporting layers between the leader and the operation. In a 50-person company there are typically two. In a 5,000-person company there are commonly six or seven. Each layer adds smoothing. By the seventh layer, in my experience, the version of reality reaching the top is an artifact of careful preparation.

• The intensity of the incentive system around reporting good news. The strongest signal a leader can read about their own information environment is what happened the last time someone surfaced an uncomfortable truth. If that person was rewarded, the information environment is healthy. If that person was sidelined, no matter how the sidelining was justified, the next person will be quieter.

C-suite leaders should take time to reflect on where these gaps may be broadest for them. For example, I worked recently with the board of a large enterprise software company that had just installed a new CEO. Within 90 days, the CEO had identified what she called “information failure points”: three functions where the gap between what was being reported and what was operationally true was, in her words, unacceptably wide. None of the leaders of those functions had done anything wrong; the structure had simply produced its predictable output.

A different version of the same problem surfaced at an AI company I advised. The founder-CEO had built a culture that loudly celebrated transparency. Every all-hands meeting opened with a candid update. The reported organization was, by design, supposed to mirror the lived organization. What I observed, however, was that the commitment to transparency felt more performative than real. While engineers would vigorously push back on one another in side conversations, they did not push back on management. The leader, who deeply wanted the truth, had no idea his organization had learned to give him a version of it that was internally consistent and externally communicable. The most polished cultures are often hiding the most divergence. The act of polishing is itself the mechanism that widens the gap.

What Leaders Can Do to Understand the Lived Organization Better

The reported organization cannot be eliminated. It exists for legitimate reasons: boards, regulators, analysts, and employees all require some form of summarized organizational reality.

The work is not to dismantle the reported organization but to govern both organizations at once. Three practices, in my experience, distinguish the CEOs and senior leaders who do this well.

The first is showing up where the work actually happens, without preparation in the room. The most effective CEOs I work with build this into recurring practice. They drop into operational reviews without their direct report present. They have small group dinners with employees three or four layers down where the agenda is unstructured. They watch what people argue about when no one is performing for them. These habits do not scale, which is the point. They are how a leader builds a different signal channel, one that bypasses the layering and the curation.

What separates the CEOs who get genuine information from those who get a curated version is also what they say and do once they are there. They ask concrete questions rather than open evaluative ones. “Walk me through what you decided this week,” produces a different conversation than “How are things?” They ask the same question of multiple people at the same level, in private, and pay attention to where the answers diverge. And they follow up visibly on what they hear. Each follow-up signals to the next person that hard truths land somewhere here. Over time, that signal becomes the strongest incentive a leader can create for people to want to tell the truth, no matter how difficult.

The second practice is making the gap itself visible. A small number of leaders I advise have built parallel measurement systems specifically designed to track the divergence. Not employee engagement surveys, which are themselves shaped by the reporting environment, but instruments designed to surface the gap. The simplest version is a recurring question to a rotating sample of mid-level managers: Name one thing your team believes that the executive team does not. The answers, aggregated and read seriously, are often more diagnostic than any dashboard.

The third is building a culture where incentives are aligned so that people want to tell the truth, no matter how difficult that truth is. Most organizations claim to do this. Very few actually do. The CEOs who succeed at it are creating the conditions that meaningfully counteract the structural pull toward a polished reported organization. They make honesty feel like the rewarded behavior, not just the right one.

I have watched the same set of practices work across very different organizational contexts. A fintech CEO set up monthly reviews where engineers were rewarded for poking holes in his strategic narrative. A financial services CEO built a discipline of opening every executive committee with a single question: What did each of you learn this week that surprised you? A large enterprise tech CRO instituted a rule that no quarterly business review could begin until each functional leader had named one thing their function was getting wrong. In each case, the gap between the reported and lived organizations narrowed because the structure had been deliberately redesigned to make the gap visible.

Leading the Reported and the Lived Organizations

The gap between the reported and lived organizations can never fully be closed. But successful leaders have learned to treat this divergence as a permanent feature of their operating environment and design governance around its existence.

The CEOs who do this best tend to share one habit: They have made it a discipline to ask, at least once a quarter, where the reported organization and the lived organization are most likely to be telling different stories. They are not always right about where the divergence is, but their willingness to assume it exists, rather than discover it during a crisis, is the practice that separates the leaders who learn from their organizations from the leaders whose organizations have learned how to manage them.

A CEO’s job is to govern both.

Copyright 2026 Harvard Business School Publishing Corporation. Distributed by The New York Times Syndicate.

Explore AAPL Membership benefits.