Build Less Data: How Focused Models Deliver More Business Value

For years, analytics teams have chased a familiar goal: completeness .

“Let’s bring everything in.”

“We might need it later.”

“Better to have it than not.”

I’ve been involved in too many data projects where this mindset quietly became the reason the solution delivered less value, not more.

The truth is uncomfortable but can also be liberating: providing less data often delivers more business impact.

“Completeness” is a Myth

A few years ago, I worked on a data warehouse that proudly claimed to be a single source of truth. It ingested data from an ERP, CRM and Forecasting application, across thousands of tables, just because it could. Six months later, business users still relied on spreadsheets!

Completeness doesn’t equal usefulness.

Most organisations only use a small fraction of available data day-to-day. The rest creates noise and results in:

• Slower development cycles.
• More complex testing.
• Confusing models.
• Reports/dashboards no one trusts.

Research repeatedly shows that analysts and decision makers primarily rely on a narrow set of KPIs aligned to their responsibilities, not every attribute available in a source system.

More data feels safer, but it rarely makes decisions clearer.

I must emphasise that this is not a call to scrap data warehouses. They will always have a place in modern data analytics. What really determines success is the layer that sits on top.

This is the layer that business users interact with every day. If it becomes bloated or difficult to navigate, value is lost regardless of how robust the underlying platform may be.

Why Minimal Models Scale Better

Smaller, focused models scale in ways bloated ones never will. Benefits include:

• Fewer tables mean faster development and testing.
• Reduced data volumes lead to quicker refreshes.
• Simpler relationships result in more predictable performance.

I’ve seen teams reduce model size by 60–70 percent, simply by stripping out unused fields and unnecessary historic data and aligning the model to real reporting needs.

In many cases, typical reporting issues like performance problems and low user adoption just disappeared, along with the volume of support tickets.

Scaling isn’t just about infrastructure. It’s about cognitive load. A model people can understand will always outperform one that looks impressive but feels intimidating.

Aligning Models to Core Workflows

The most effective data models don’t mirror systems; they mirror how the business works.

Finance teams care about month end close, forecast accuracy, and variance analysis. Operations teams care about throughput, backlog, and utilisation. When models are built around these workflows, data becomes embedded in decision making rather than sitting in dashboards.

One of the most impactful shifts I’ve seen is moving from “What data do we have?” to “What decisions are we trying to support?”

The latter consistently drives more disciplined and valuable data modelling choices.

Saying “No” as a Modelling Strategy

This might be the hardest lesson for any Data Engineer or Consultant: good data modelling requires saying “no.”

“No” to speculative requirements.

“No” to dumping raw schemas into production.

“No” to the idea that we must solver every future question.

Saying no isn’t blocking value, it is protecting it. Each exclusion keeps the model lean, readable, and aligned to what matters now.

“Excluding” data from the model does not mean it is unavailable in the underlying data warehouse. It is simply hidden from the data model, using views. When new reporting needs arise, they can be added intentionally, not accidentally.

Saying no early prevents unnecessary complexity later.

The ROI of Building Less

Focused models deliver tangible returns:

• Smaller data volumes reduce development and infrastructure costs.
• Faster delivery means quicker time to value.
• Clearer models ensure the business gets exactly what they need, nothing more, nothing less.

In analytics, restraint is a competitive advantage.

If your data platform feels heavy, slow, or underutilised, the problem may not be what you’re missing, it may be what you’ve built too much of.

Sometimes, the smartest thing a data team can do is build less.

Next Steps

If your analytics environment feels heavier and harder to use than it should be, the issue is often not the platform itself but how data has been engineered and modelled. At 5Y, we help organisations simplify their analytics foundations by standardising data through our Engineering Hub, then delivering focused, decision led analytics through our Transformation Hub and out of the box modules aligned to real business workflows.

This approach allows teams to stop over building and start delivering analytics that genuinely drive value. If you want to explore how this could work for your organisation, click on Contact Us and we will be in touch to discuss the best next steps.