When working with data, it’s tempting to think accuracy is everything. After all, who doesn’t want to be right? But in practice, there’s an important difference between data that is correct and data that is useful.
Take the classic example: How tall is Mount Everest?
The “official” number is about 8,849 meters (about 29,032 feet). But here’s the catch: that measurement depends on sea level, which itself isn’t fixed—it rises and falls depending on tides, climate change, and local geography. Factor in tectonic shifts, snowpack, or even the precision of the instruments, and you could argue the number is never perfectly correct.
Does that mean the common number is wrong? Not at all. For 99.9% of people, 8,849 meters is both good enough and useful. Only specialized researchers or engineers need the decimals, the margin of error, or the changing sea-level context.
This brings us to the old saying: “Perfect is the enemy of good.”
Chasing perfect correctness often slows down decision-making, confuses audiences, or diverts energy from what really matters. The pursuit of perfection can paralyze projects, leaving teams stuck in endless refinements instead of delivering something that helps people act.
Useful data answers the question at hand with enough reliability to guide action. Correct data tries to represent an absolute truth—something that’s often slippery, context-dependent, and in some cases unknowable.
The key is to ask: What level of correctness is necessary for this purpose? If you’re setting a record in the Guinness Book, millimetres might matter. If you’re wondering about the world’s highest mountain, the rounded figure works just fine.
And if you think “good enought” is a radical notion, consider the world of finance. There is an important financial concept of “materiality” that helps finance folks avoid the “perfect” trap.
Most data we work with is ultimately for managerial decision-making or for some aspect of operations management. And while we sometimes care about high precision and statistical validity, most cases simply don’t need that.
So, in short: don’t let the quest for “perfect” data get in the way of providing useful answers. Precision has value, but in its place and only in proportion to the decisions it informs.







