Everything in life leaves a trace. As we mentioned in the previous chapter, most events and phenomena can be read in advance, interpreted, and acted upon accordingly. In the Growth phase, the entire purpose of the work people and organizations do on themselves — the experience, the perspective, the gains — is to reach the competence needed to do exactly that.
But in the Setup phase, we have to move into another dimension of it: Measurability and Data Reading. Naturally, the information, data, and headline metrics you need will vary depending on your sector, your line of business, your country, your customers, and even the way you operate. But at the Setup stage, every person and every organization must decide which data to measure, which methods to use, and which sources to trust. Whether or not to build on data is not a matter of preference.
In the many companies I've consulted for, worked with, or simply come to know, I keep seeing two distinct, wrong approaches. The first is something I call data intoxication. The second is a failure to separate the person from the institution. Let's take each in turn.
Data intoxication is what happens when organizations or individuals track data that doesn't fit their actual circumstances, or read their data with an overly optimistic eye. We see this especially among company owners, marketing departments, and sales teams. A wholesale brand that barely calculates unit-level profit and instead leans on periodic profitability is a good example. So is a marketing team presenting reach and impressions in reports — not because those numbers matter, but because they make the team look more successful or help justify a bigger budget. So is a salesperson logging every phone call as a qualified lead. In countless press releases and public statements, we see hollow figures — "we grew our volume by this much this year," "our team grew by this many people" — presented in the language of success, even though they contribute nothing to profitability, efficiency, or actual results.
This is deeply dangerous for both individuals and organizations. It doesn't just blind them to their pain points and weaknesses — it causes them to make even more mistakes. Ego, avoidance psychology, self-interest: plenty of forces are at play here, and all of them are personal factors. Which brings us to the second common problem.
The failure to separate person from institution shows up not just in data, but across many other points as well — and it directly overlaps with something we covered earlier: the fact that people are the fundamental asset. We'll return to this in later chapters. But where it creates trouble specifically around data is this: a person's desire to appear successful, or their department's or title's own interests, poisons the data itself. Combined with the examples above, we often see that the right data and the results that actually matter get pushed aside by people, in favor of what serves them personally. In small structures, this may not cause collective damage. But in large organizations, this is precisely why national and international oversight bodies exist. Departments that withhold necessary data because it doesn't flatter them, cold wars fought over titles, owners who can't separate company revenue from EBITDA and treat the difference as their own pocket money, subcontractors chasing smooth-talking schemes — these are just a few examples of this spreading dysfunction.
Numbers and mathematics cannot lie. People, however, are a different story.
So how can people and organizations actually build a data-driven structure? There are four fundamental pillars.
The first is field data. People and organizations must fully understand the core data and measurement metrics of the field they operate or were founded in, and accept the ones that serve their purpose as their primary guideposts.
The second is data sources. This may be the single most important pillar of a data-driven setup. A wrong data source, an inconsistent channel, an unreliable survey result — any of these can render everything built on top of it wrong, and the damage can be severe. That's why the most accurate and reliable sources must be chosen. The most unshakeable data, in terms of accuracy, is our own data. Which brings us to the third pillar.
The third is data infrastructure. People and organizations must collect data. How many customers did we talk to? What are the main complaints? What are the reasons for returns? What's the ratio in our team's performance measurement? What are our core bottlenecks? The metrics for these questions, how they'll be measured, and how they'll be audited must all be established at the Setup stage. The most reliable data is the data that comes from our own measurement.
The fourth is data validation. Because of the issues we've described above — the failure to separate person from institution, or data intoxication — it's essential to have reliable independent auditors, and eventually a dedicated internal audit function as well. At the same time, the founder — the person at the top — must know exactly what needs to be measured in each domain, even domains outside their own expertise. Because, as we always say, people are the most important asset, and continuous growth and learning are essential for everyone.
So far, we've explained why we need to live by data and lay this into our foundations during the Setup phase. But are there real-world examples of this? As always, we've compiled the core real-life cases and details for this chapter:
The most instructive example of Data Intoxication is, once again, WeWork and its founder Adam Neumann. In its investor materials, the company invented a concept called "Community Adjusted EBITDA" — stripping out rent expenses, marketing costs, even general corporate overhead, to construct a narrative that made the company look far more profitable than it was. This is close to a textbook case of failing to separate company revenue from EBITDA; Neumann's relationship with data wasn't about measuring, it was about persuading. When the IPO attempt collapsed, the gap between reality and what had been reported wiped tens of billions of dollars off the company's value within days.
The most striking example of the failure to separate person from institution is the Wells Fargo scandal. Bank leadership set impossible cross-selling targets for employees; thousands of staff, unwilling to lose their jobs, opened roughly two million fake accounts under this pressure, without customer knowledge. Here, the data poisoning came directly from personal interest — from anxiety over title and job security. CEO John Stumpf presented those "successful" cross-selling numbers as a point of pride for years, because questioning the reality behind the figures served no one's interest. Because the institution couldn't separate its own employees' personal motivations from the data, its most basic performance indicator was wrong from top to bottom.
For the power of choosing the right field data, the best example remains Oakland Athletics and general manager Billy Beane. In the 2002 season, he set aside metrics like batting average — sacred to baseball for decades — and turned instead to on-base percentage, a stat almost no one valued at the time. With a fraction of his rivals' budget, he carried a team to one of the longest winning streaks in league history, simply by choosing the right field data — not the data everyone else was watching, but the data that fit the purpose.
When it comes to the reliability of a data source, the Boeing 737 MAX disaster is an unforgettable warning. The aircraft's MCAS system based its critical decision to push the nose down automatically on data from a single angle-of-attack sensor — with no redundant, cross-checked source structure. When that one sensor failed, the system treated the wrong data as correct and activated, leading to the crashes of the Lion Air and Ethiopian Airlines flights. It is the most costly example of how a wrong, inconsistent, or single-point data source can corrupt everything built on top of it.
The most mature example of turning data infrastructure into a discipline is Amazon. At the center of Jeff Bezos's corporate culture sits the weekly business review — meetings where hundreds of the company's operational metrics are gone through one by one. Known for banning PowerPoint decks in favor of six-page narrative memos, Bezos requires every one of those memos to be grounded in data the company measures itself. This self-collecting reflex, institutionalized over the years, shows how a culture built on measurement rather than guesswork actually gets constructed.
The absence of data validation finds its most striking form in Theranos. Elizabeth Holmes's company claimed it could run hundreds of tests from a single drop of blood — but no data supporting that claim ever passed independent audit. Engineers inside the company knew the technology didn't work, yet the founder's charisma and narrative kept the unverified data standing for years. The structure that should have existed for independent auditors was exactly the step Theranos skipped — and that omission brought a multi-billion-dollar company to zero overnight.
