Thirty Years of Excel Finally Over

Ontology is an old word for something simple: the study of what exists and how it connects.

Aristotle coined it over two thousand years ago. His question was not: What do we know? But: How does what we know connect? You can collect a thousand details about a market and still not understand how it works. Who buys what, where, why, and what does regulation have to do with it? The details alone do not answer that. Only when you build the relationships between them does something usable emerge.

An example. We know that daylight improves concentration. We know that noise reduces it. Two studies, two results. But when you are planning an office that is bright and loud, the studies alone do not help. You need the relationship between them. Which effect is stronger? At what noise level? For what kind of work? Does it depend on age, time of day, whether someone sits alone or in a team? The answers are somewhere in the research. But they are not in individual studies. They emerge when you build the structure between them. That is ontology.

Why I am writing about this

I learned the word ontology only a few years ago. But what it describes, I have been doing for thirty years.

I currently advise American companies entering Europe. They come with data about their home market and their customers. Europe has different regulations, different buying habits and different distribution channels. The question is always the same: What transfers? What does not? And what does it depend on?

Ten years ago you answered that with research, spreadsheets and experience. Today it only works through data, and AI handles a large part of the work. The volume of information has become too high for manual processing. So I build systems for it. Not as an experiment but as a tool that is usable and transferable. The structure behind these systems is ontology.

And it makes the work better. Not because the machine is smarter. But because I am forced to explicitly build the structure that used to exist only in my head. When you have to define relationships between things properly, you quickly notice where you relied on gut feeling before and where your own assumptions were wrong.

It is more fun too. You understand better what you are actually doing.

Looking back

I introduced and distributed Power Balance across Europe. I imported health products from Japan. I advise US companies entering the European market and build logistics and distribution structures for them, direct or retail, depending on what makes sense. The content was different every time. My approach was the same every time.

In every case, market data and sales and marketing strategies were decisive. And there was always an enormous amount of data. Manually researched, processed, understood and communicated across teams. An immense effort and you were always behind. Nothing was in real time.

Databases and algorithms got better over the years, the work got faster. But the volume of spreadsheets and data sources was something for nerds, not for people in the market. It was clunky everywhere. The different customer systems of large retail chains, the interfaces, the data formats. Even small distributors needed significant IT effort just to keep up with daily operations. Data, content and APIs dominated everything and the smarter you handled them, the faster and more effective you were. Permanent monitoring and controlling were mandatory.

But who actually understood the data? I cannot answer that because I struggled with it myself. What I can say: not everyone understood it. Most people ignored it and went with their own interpretations. The successful ones were those with the highest hit rate. But there was a lot of luck involved and a lot of coincidence. A good manager or sales person was the one with the right instinct, the feel for things. That is true. But even the best noses had bad luck sometimes and got demoted. Often unfairly.

I had no word for my approach. I called it “my system” or “my method.” It was a kind of ontology, lots of Excel, little elegance. The attempt to make connections visible in all that chaos, connections that someone could base a decision on.

With today’s systems, more people understand what is in the data. Not everyone, but significantly more than before. That changes the work.

I switched sides. From player to coach. Instead of fighting in the market myself, I now build the systems that others work with. That means I can go deeper across a broader range than before. It is fun. The urge to explore.

Palantir

Palantir is how I became aware of ontology. I am not a scientist and would probably never have come across the term otherwise.

Palantir was founded in 2003 and initially worked for intelligence agencies and the military. Their problem: enormous amounts of data from different sources and the task of finding connections in them. Palantir’s solution was to build their data layer as an ontology. They call it exactly that. In their system Foundry, “the ontology” is the core: a model that defines what things exist, what relationships they have and what properties those relationships carry. The connection between two data points is its own object with its own attributes.

Ontology existed before Palantir in bioinformatics, in the semantic web, in library science. But Palantir made it commercially visible. Since they connected their ontology with language models in 2023, Fortune 500 companies talk about “building their ontology.” That is how I came across the word.

What comes next

A question is on my mind: Is what goes by the name “agentic AI” essentially ontology? A system that breaks down tasks, recognizes dependencies and executes steps in the right order. To me that sounds like ontology with a new label. But I need to understand this better.

Why AI without structure does not get you far and what ontology changes about that is what I will write about in the next essay.