The Hidden Order of Chaos: Why You Need Less Data Than You Think

In my previous post, "When Measurement Makes You Worse," we explored the dangerous trap of "vanity metrics"—precise numbers that create false certainty while failing to inform actual decisions. But if the problem isn't a lack of data, what is the solution?

The answer lies in understanding a beautiful, almost mystical feature of our natural world: The Law of Large Numbers. By mastering this principle, you’ll realise two things that will change how you lead: you need far less data than you think, and you already have more data than you realize.

The Supreme Law of Unreason

Francis Galton, a pioneer of modern statistics, once described the orderly way that patterns emerge from chaos as a "cosmic order". He was captivated by the Law of Large Numbers, which states that as you collect more independent observations, their average will inevitably converge toward the true population mean.

Think of a coin flip. A few flips might look like a random streak of heads, but keep flipping, and the "noise" of those early imbalances is eventually overwhelmed by the sheer volume of new, independent events. In your organisation, this means that while individual projects or customer interactions may seem unpredictable, the aggregate behaviour of your "system" is governed by a beautiful, predictable regularity.

The Rule of Five: The Leader's "Mathless" Power Tool

If the Law of Large Numbers tells us that data eventually converges, Hubbard’s Rule of Five tells us how quickly we can start making meaningful inferences.

Many leaders suffer from the "uniqueness fallacy"—the belief that because their project is complex, they need thousands of data points to measure anything. Measurement science proves the opposite: if you know almost nothing, almost anything will tell you something.

The Rule of Five states that there is a 93.75% chance that the median of an entire population falls between the smallest and largest values in a random sample of just five.

The Practical Example: If you are uncertain about how long a recurring task takes, randomly pick five past instances. If the shortest was 2 days and the longest was 12, you are now over 90% certain that the real median time for that task is between 2 and 12 days.

If that range is narrower than your previous "gut feel," you have successfully reduced uncertainty. That is a measurement.

You Have More Data Than You Think

We often hear teams lament, "We don't have enough data to be statistical." In reality, organisations are usually swimming in "found data"—administrative records, old project logs, and "trails" left behind by every business process.

The sources offer a contrarian set of assumptions for leaders:

    It has been measured before: No matter how "unique" your problem feels, someone has likely researched and published data on it.

    You have more data than you think: If you look for "traces" (like test logs or email timestamps), you'll find data you've been ignoring.

You need less data than you think: Because the largest reduction in uncertainty happens with the first few observations, you rarely need a "big data" set to change a decision.

From Discovery to Decision

The goal of leveraging the Law of Large Numbers isn't to build "binders full of measurements" for curiosity's sake. As we established in the last post, a measurement has value only if it reduces uncertainty for a specific decision.

Stop asking, "What can we track?" and start asking, "What is the most uncertain variable that could change our next big move?". Whether you are sampling five items or five thousand, the math of the universe is on your side—ready to help you find the signal in the noise.

--------------------------------------------------------------------------------

You don't need "Big Data" to be a scientific leader. You need a Decision-Driven mindset and the courage to trust that even small samples reduce the risks of guessing