Reading Time: 5 minutes
When we talk about data architecture — whether it’s in support of analytics, AI, data science, or combinations of these and other initiatives — we usually start with technology: what’s new, what’s modern, what’s changing. But in this post, I want to start somewhere else: with the organization itself. Because technology should support the way people work, and not the other way around. To explain why that matters, I want to start with a quick story.
At an organization I worked for previously, leadership would set the annual goals, which often revolved around broad concepts like growth, innovation, and efficiency.
Those goals intentionally flowed down through the organization to create alignment to the company’s strategic direction and each team had a different interpretation of what those goals meant. Finance saw one version of “growth,” Sales another, Operations and IT others still. Even though we were aligned in intent, each group envisioned a different path forward so as goals moved closer to the day-to-day work, they naturally became more concrete. What started as high-level intent turned into specific targets, decisions, and actions for individual teams and people. This was how strategy moved from vision to execution.
This is where the role of IT becomes critical. How IT understands the organization — how people work, what questions they ask, and what success looks like in each domain — directly shapes the architecture we put in place. If IT sees the organization narrowly, the architecture will be narrow. If IT sees the bigger picture, the architecture can truly support the enterprise in all its complexity. The big question becomes: How can we build a data architecture that supports everyone, and not just a select few?
The Diversity of Innovation
Every team works differently. Finance cares about forecasting. Sales cares about revenue and product mix. Manufacturing cares about quality and uptime. Marketing cares about reach and attribution. Even within the same department, people ask different questions in different ways. Those differences already shape how people use data. When AI enters the picture, they show up even more clearly: different groups interact with AI in different ways, using their own language, assumptions, and definitions. This diversity isn’t a problem — it’s what drives innovation.
The problem is when our architecture assumes that everyone works in the same way. If we want such initiatives as analytics, data science, and AI to support the whole business, the architecture has to accommodate this diversity — not flatten it.
Where Traditional Approaches Break Down
This is where the friction appears. Traditional data management processes were built around large, well-defined projects — the few initiatives that get formally funded and resourced. But most people across the organization don’t have funded projects. They just have goals they’re accountable for. So they find their own ways to work: spreadsheets, extracts, side databases, whatever helps them meet their objectives. This isn’t rebellion, it’s a necessity. The formal system doesn’t support the pace or the variety of the many kinds of questions people need answered. The result? Still many small, disconnected solutions that work locally… but create inconsistency across the enterprise.
What People Actually Need
After seeing this pattern repeat across a variety of different industries, the core needs are clear: Most people don’t have “data projects”; they have questions and decisions.
To answer them, they need:
- Reliable access to data
- One place to start — not ten
- A consistent, intuitive experience
- Freedom to explore without waiting in line for IT support
- Confidence that they’re operating within the right controls
- Information that’s understandable, even if they’re not a domain expert
And if they can’t get that through official channels, they will find alternatives. So the real challenge is: How do we provide broad, simple, intuitive access while maintaining the data governance and trust that the organization requires?
The “People-First” Data Architecture Test
To answer that, there are seven practical questions I think every architecture should address, in order to build a “people-first” data architecture, or an architecture that provides people with what they actually need.
If you can answer Yes to the following seven questions, then you have a people-first data architecture:
- Does it support the company’s goals? Not just technical goals or the goals of specific, individual initiatives — everyone’s goals.
- Is it simple enough for non-experts to leverage? People shouldn’t need to understand the underlying systems.
- Is it consistent — is there one place to go? Access can’t be scattered across tools and silos.
- Is it flexible enough for exploration? New questions shouldn’t always require IT projects.
- Is it secure-by-design? Security and data governance must be centralized at the enterprise level.
- Can IT support the architecture without becoming a bottleneck? Architecture should scale with demand, not load IT with every request.
- Will it age well, as the organization evolves? Organizations change faster than platforms, and the architecture must be able to keep up.
Why Logical Data Management Matters
I’ve talked about the limitations of traditional data management approaches. Fortunately, logical data management provides a more practical alternative — one that focuses on making information usable across the organization, not just technically accessible. Instead of forcing people to navigate systems, schemas, or locations, it enables data to be packaged as meaningful, organization-aligned representations and relationships that people can actually act on.
By introducing a logical layer, organizations create a shared way of working with information that reflects how the business thinks and operates. People don’t need to know where data lives or how it’s physically managed. They interact with information expressed in familiar business terms, and the logical layer resolves what’s needed at the moment it’s used. This makes the experience consistent whether someone works with data every day or only when a question arises.
Because the logical layer spans the enterprise, it also becomes the natural place to apply security and governance once, and have those controls follow the information wherever it’s used. At the same time, it enables information to be shaped differently for different patterns of activity — from highly technical users building complex analyses to occasional users answering simple, time-sensitive questions — without duplicating data or experiencing disruption from underlying systems. Logical data management scales not just technically, but organizationally.
In this way, logical data management reflects how organizations actually function. People move fast. Questions emerge unexpectedly. And data lives everywhere. Instead of forcing the business to adapt to the architecture, logical data management lets the architecture adapt to the business.
When we look back at the seven questions, the picture becomes clear:
- Traditional approaches, including data warehouses and lakehouses, answer technical needs well — scaling storage and processing.
- But they struggle with organizational needs — broad access, simplicity, consistency, exploration, and centralized governance.
A logical data management layer sits above the lakehouse and other systems to close those gaps:
- It gives people one consistent way to access information.
- It makes the experience intuitive for non-experts.
- It centralizes policy across many systems.
- It reduces the operational load on IT.
- It adapts naturally as new teams, tools, and questions appear.
It supports the whole organization — not just the parts with technical projects. And it works with complex data in some of the largest companies in the world.
The Big Picture
In the end, analytics, AI, and data science are just tools. What organizations really need is a foundation that lets people think, explore, and innovate using trusted information, and that starts with people. When we understand how people work, how they ask questions, and how they use data to meet their goals, we can design an architecture that truly supports them. Solve that problem first, and the rest will fall into place. Logical data management provides that foundation. It aligns data architecture with how people actually operate. It supports the full spectrum of goals across the enterprise. And it helps the organization see — and use — the bigger picture.
