There is an old saying that goes, “Conviction can be worse than madness.”
Sometimes we become so convinced that something is true that we defend it at all costs—only to discover that the numbers tell a completely different story.
So let me start with a point we have made several times before:
It is impossible to talk about continuous improvement in a company without collecting and analyzing data. Otherwise, there is a real risk of making decisions based on assumptions and personal beliefs rather than objective evidence.
That said, I would like to add another important consideration.
While it is true that data collection—both in business and elsewhere—undeniably improves performance and helps reduce costs, and
while data is objective, impartial, and neutral, it is equally important to recognize that, within a given context, the conclusions it leads to depend on how it is interpreted.
Let me give you a practical example.
Imagine a company that manufactures mechanical components and has a scrap rate (the percentage of non-conforming parts) of 1%. On average, that would generally be considered a very good result.
However, if you dig deeper and break the data down by department, you might discover that Department A is performing quite well, while Department B is generating the vast majority of the scrap, making it almost entirely responsible for the company’s overall rejection rate.
That same 1%, which initially appeared to be an excellent result, remains low when viewed at company level. But when analyzed within the specific context of Department B, it reveals considerable room for improvement.
So, returning to the original question:
Do data really tell us everything? Yes—but only if they are interpreted within the proper context. In other words, data must be broken down and analyzed using models specifically designed to take relevant variables into account.
So, how do you build a model that actually works?
First and foremost, by applying logic and sound judgment. To this, you must add a clear mission—the objective assigned to each department—so that people understand not only what they are expected to achieve, but why. This enables them to work according to a roadmap built on a holistic view of the company.
For example, one department’s mission might prioritize quality over production speed, while another might prioritize throughput. Neither approach is inherently right or wrong—the important thing is that the objective is clearly defined from the outset.
Another crucial factor that should never be overlooked is the expertise of the people responsible for designing and interpreting these analytical models.
No matter how skilled a data analyst may be, they must also possess a solid understanding of the specific field they are working in.
In our case, that field is mechanical engineering and precision gear manufacturing.