It’s hard to believe today, but there was a time when people only collected data if they absolutely had to. The stereotypical images of the 1970s office, with rows of filing cabinets and index cards, reflected a very different attitude toward data. You kept what you absolutely and positively knew you were going to have to refer to – and nothing else.
At that time, anything beyond a company’s basic data was considered business waste. Data was a byproduct, not an asset. This was largely driven by technology. Even when we moved from paper to digital, digital storage was slow, expensive, and difficult to operate and analyze. Even if the data was recorded, it was often considered write-only, saved but never mentioned. Data was a liability: expensive to store and even potentially dangerous.
However, as technology evolved and analysis techniques developed, things changed. Over the past two decades, the way we view the data we generate and collect has constantly evolved. From commercial exhaustion, it quickly became an essential marketing and commercial asset: the new oil, as we have often been told.
Your customers are searching everywhere. Make sure your brand introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
Start free trial
Start with
How data became the center of marketing
This shift has caused companies to rethink what data they collect and why. Even if you didn’t know how to use it, it became imperative to store all data, even the finest transactional data. Data management technologies and techniques evolved such that data lakes, pools and oceans arose, and all data was now clean and available for analysis. In theory, at least.
As our analytics and data science capabilities grew, we moved from a descriptive approach (“What did the customer buy?”) to a predictive approach (“What are they likely to buy next?”). This type of information is extremely valuable to a business because it allows us to evolve our offerings and operations to meet consumer demands and optimize performance.
But there was still one step to take: moving from predictive to prescriptive. This step goes beyond simply indicating what the customer is likely to do next and instead indicates what we should do next. Systems began to arise that gave us the next best action – what we should actually be doing. For the most part, they were relatively limited in scope (i.e. what deal to offer next or what discount to apply), but nonetheless provided us with a powerful way to adapt to ever-changing customer and market demands. This is all based on the data we collect.
All of the above relies on us treating data as the asset we come back to. The goal of more advanced analytics – whether descriptive, predictive or prescriptive – is to give us a better view of the data we have and what it means for our business.
Why AI models are changing the role of data
We now see ourselves facing another major technological shift, as LLMs and other AI-related technologies radically change the way we work. It can be tempting to think of these new approaches and technologies as simply better ways of working with the data we have – and in some ways, that is the case. However, if you step back and ask what role data plays in these technologies, you’ll see that it’s much more radical than just cool new tools.
To understand this, you have to look a little under the hood. The majority of modern LLMs are built on an architecture called transformers. They take your input text and process it using billions of parameters (mathematical rules) drawn from a massive initial diet of data. The way they store this knowledge can be simplistically thought of as file compression.
The text “What is the capital of France?” successfully generates “Paris” not because the model contains a search engine, but because its parameters effectively act as a compressed and lossy recall of the entire original training set. Although imperfect, this analogy is useful. As science fiction author Ted Chiang said, an LLM is like a “Blurry JPEG from the web.”
The implication is that once a model has been trained, it contains all the knowledge it will retain (at different levels of fidelity). When we use a model, we do not go to the source, but to an imperfect snapshot of it. If you think of the fuzzy JPEG analogy, our challenge is to complete the model with a clear, high-definition image of our business, derived from our own proprietary data.
Because the scope of today’s base models is now so broad, they are excellent at the prescriptive part of the workflow, not only analyzing but indicating what we will do next. With your own data asset, you now have the capability we’ve been working on: moving directly from data to action.
What this change means for your data strategy
One technology that is contributing to this shift in how we use data is the Model Context Protocol (MCP) – a standardized way of exposing our proprietary data to models – becoming the universal adapter that allows models to read your database live without swallowing it permanently into their fuzzy memory. MCP is still in its infancy and probably won’t be the definitive form of interaction between data and models, but it shows how necessary it is becoming to rethink the role of our data assets.
This means we now need to rethink the role of our data. If the primary purpose of our data is either to train or complete a model, does that change what we collect and when? Does this change its value and role in our marketing and sales landscape?
The challenge today for everyone who collects business data, and surely we all are, is how do we change our thinking to recognize that data is no longer the central asset? Companies that radically rethink the role of their data assets will thrive in this new ecosystem.
Key takeaways
- Data has evolved from a stored asset to something that powers and shapes AI-driven decisions.
- The evolution from descriptive analytics to predictive analytics to prescriptive analytics paved the way for today’s AI workflows.
- Large language models do not retrieve data in real time, they rely on compressed knowledge that must be supplemented with proprietary data.
- The real advantage now lies in combining basic models with high-quality business data.
- Marketers need to rethink their data strategy, from collecting all the data to using it for models and real-time decision-making.
AI-generated list
