We all want data — but we mean different things. Here’s what that really means for AI, insight, and your competitive advantage.

Every company says they want to be “data-driven.”

Every leader says “we need better data.”

Every team wants to use AI.

And yet… something is hard behind the scenes.

Dashboards are good but they can’t show the whole picture so let’s extract to excel.

People argue over definitions thinking everyone means the same thing as they do!

Not because anyone is bad at their job, but because they’re all using the same word, DATA, to describe completely different things.

Let’s fix that!

1. The four types of data — and the real Reason for all the confusion

Think of your company as a factory that constantly produces information.

That information moves through four stages — and confusion happens when people mistake one stage for another.

1. Operational Data — the raw stuff created in systems as things happen

This is what your systems automatically generate every day:

Orders placed, payments processed, applications submitted, logins recorded.

It’s sometimes called raw data because it hasn’t been cleaned or translated yet.

If you’ve ever looked into a system log or database and seen cryptic field names like cust_id, txn_amt, or flag_1, you’ve seen operational data.

Developers created it to make systems run, not to make sense to humans.

Operational data tells you what happened, but not always in words the business understands.

It’s factual — but it’s like a language no one taught you how to read.

2. Analytical Data — the version reshaped for human questions

This is the data analysts and BI tools work with: cleaned, joined, and organized so you can ask questions like:

• How many customers did we gain this month?

• What’s our average order value?

• Which region is performing best?

Analytical data takes the raw noise from systems and turns it into patterns people can discuss.

It’s what appears in dashboards, reports, and Power BI charts.

The catch? Every dashboard reflects someone’s definition of reality.

If teams don’t share the same definitions underneath, you’ll end up with multiple “truths” that all look right — in isolation.

3. Derived Data — the predictions and scores created from other data

This is where AI and analytics meet.

Derived data isn’t something your systems record — it’s something your models calculate on top of your systems in your analysis tools or god forbid in excel or local calculations:

• Churn probability

• Credit risk

• Customer lifetime value

• Forecasted demand

Derived data tells you what might happen next.

It’s smart — but it’s also synthetic.

It inherits every assumption and mistake from the layers below it.

If your operational or analytical data is off, your derived data just becomes confidently wrong.

4. Semantic Data — the hidden layer that gives everything meaning

This is the most powerful and unfortunately most ignored type of data. But forget real AI without it.

Semantic data defines the language your company uses to describe itself.

It answers questions like:

• What exactly counts as an “active customer”?

• How do we measure “revenue” across products and markets?

• Which systems hold the source of truth for each concept?

Semantic data isn’t a dataset — it’s a dictionary for your business.

It’s what allows your AI, analysts, and leaders to all understand the same question in the same way.

Without this layer, you don’t have one truth — you have a hundred local ones.

NOTE: This is the hardest part, because getting people aligned and changing the culture, ways of working, defining every detail of your business and following it throughout the company.. that’s the real transformation!

2. Why business people love Excel (and always will)

There’s a reason people still live in Excel.

It’s not stubbornness.

It’s because Excel feels safe.

You can open it, see the numbers, tweak the formula, and instantly watch the result change.

You don’t need to wait for a data team or learn a new tool.

You’re in control.

Excel feels like reality.

It mirrors how people think:

“What if I change this?” → “What happens to that?”

It’s not perfect — but it’s human.

3. And aow… Excel has an AI Co-Pilot

Anthropic recently introduced Claude for Excel — an AI assistant that works inside Excel.

You can ask it to summarize data, explain formulas, fix errors, or even build templates for you.

It’s fast, context-aware, and surprisingly capable.

You type a question in plain English — Claude reads your spreadsheet and answers right there.

For anyone who already lives in Excel, that’s magic.

But it’s also a potential trap.

4. The trap of local truths

If your company starts building AI on top of Excel files, each workbook becomes its own tiny universe.

Different definitions.

Different assumptions.

Different rules.

You’ll move faster — but only inside each silo.

Soon, you’ll have 50 smart spreadsheets and zero shared understanding.

That’s not digital transformation — it’s automated fragmentation.

We have seen it allready in many companies, hundreds of dashboards and excel sheets people trust. Then they fight if they are correct and eventually blame IT for mess they built not understanding how data works.

5. The real future: AI built on your Data Platform

Now imagine a different setup.

All your operational, analytical, and derived data lives inside a governed data platform — a single ecosystem designed to make sense, not just store.

And on top of it sits a semantic layer: your shared business language encoded in code.

Now your AI, dashboards, and even Excel sheets all plug into that same semantic foundation.

When someone asks,

“What’s our active customer base this quarter?”

the system knows exactly what “active” means — because it’s defined once, reused everywhere.

That’s how you move from data chaos to data clarity.

Ofcourse not everything needs to be here, but those most important things that drive the business forward, tailored for meeting business goals and compliance requirements.

6. The difference in plain language

AI in Excel gives you speed.

AI on your data platform gives you scale.

AI in Excel helps one person solve a local problem.

AI on your data platform helps the entire company speak the same truth.

When every tool, team, and algorithm draws from the same semantic foundation, you stop arguing about numbers — and start improving them.

Then those LLMs can be used everywhere, from customer service, business decisions, hypothetical predictions and ofcourse in products.

7. The next step — From local AI wins to company-wide intelligence

It’s tempting to think: “If AI already works in Excel, isn’t that enough?”

And yes — it’s a great start.

Every person experimenting with AI inside Excel is showing you what’s possible.

They’re identifying real pain points.

They’re discovering how AI fits into their workflow.

But if you stop there, you’ll never move beyond AI islands — smart experiments that can’t talk to each other.

The real opportunity is to connect those local wins to your central data foundation.

That means:

• Capturing the best logic, prompts, and formulas emerging from those Excel pilots.

• Turning them into governed metrics, shared definitions, and reusable templates.

• Rebuilding them into your data platform so every team benefits from what one team learned.

That’s how you turn AI in Excel into AI across the enterprise.

The goal isn’t to stop people from using Excel — it’s to connect it to the same foundation your AI relies on tomorrow.

Because when every local insight strengthens a shared ecosystem, your intelligence compounds.

Final thought

AI in Excel is a glimpse of what’s possible.

AI on your data platform is what makes it sustainable.

Let local innovation continue — but rebuild the best ideas centrally, with shared definitions and semantic context.

That’s how you turn individual efficiency into organizational intelligence.

That’s how you move from doing data to understanding it.

And that’s how you build an AI strategy that actually scales.

//Adnan