Why Your Data ROI Is Under Scrutiny (And What to Do About It)
The post-ZIRP reckoning for data departments
Tony Zeljkovic
2026-03-10
Many data-heavy organizations are heavyweights in data analytics but are severely underweight in delivering a return on investment (ROI) from their data.
With the unwinding of decades of zero-interest funding, the ROI of data departments across industries is under heavy scrutiny. The modern data stack, fueled by this environment, has driven rapid growth in business intelligence capabilities for many companies — often accompanied by rapidly increasing expenses on cloud data warehouses connected to BI platforms.
There is a growing desire for more value-added activities beyond delivering dashboards within an increasing number of organizations. An increasing number of companies are relying on in-house solutions to save millions of dollars in costs while doing so.
The Dashboard Trap
Most data teams start their journey by building dashboards. It's a natural first step — stakeholders want visibility, and BI tools make it straightforward to deliver.
But here's the problem: dashboards are passive. They show you what happened, but they don't do anything about it. As your data maturity grows, the gap between "we can see the problem" and "we can fix the problem" becomes the biggest bottleneck.
Higher management increasingly perceives additional investments in BI platforms as yielding diminishing returns on investment. And they're not wrong.
The 80/20 of Domain Expertise
Here's an uncomfortable truth for technical professionals:
Roughly 80% of your perceived value comes from domain expertise — not technical skill. Your understanding of how the business works, what drives revenue, what creates costs, what introduces risk. The remaining 20% is execution.
Many businesses don't care about tech debt, software architecture, or engineering best practices. They care about growing the business. Whether that's a healthy perspective is beside the point — it's how buyers evaluate what you're selling.
A C-suite executive with deep knowledge of where the business was bleeding money used AI tools to build a process automation application. The architecture might be questionable to a purist. But it saved tens of millions of dollars because this person knew exactly which problem to solve.
Meanwhile, a senior architect spent months building a sophisticated infrastructure management system. Clean code, proper patterns, impressive complexity. The reception was polite but lukewarm.
The executive knew what needed to be done. The senior engineer did not.
Beyond BI: The Data Application Layer
The next evolution is building data applications — interactive tools that let business users take action on data insights directly. Think of it as the difference between a weather report and an autopilot system.
What Makes a Good Data Application?
- Action-oriented — Users can make decisions and trigger workflows, not just observe
- Domain-specific — Built for a particular business process, not a generic dashboard
- Self-service — Business users can operate independently without data team intervention
The Technology Stack
We've found that the most effective approach combines:
- Python-based frameworks (Streamlit, Dash) for rapid prototyping
- Container orchestration (Kubernetes) for scalable deployment
- Cloud-native services for caching, authentication, and storage
This stack lets data engineers — who already know Python — build production-grade applications without needing to learn traditional web development.
Measuring What Matters
The key metrics to track:
| Metric | What It Tells You |
|---|---|
| Time to decision | How fast can users act on insights? |
| Self-service ratio | What % of data needs are met without data team involvement? |
| Cost per application | How much does it cost to build and maintain each app? |
| Revenue impact | Can you tie data work directly to business outcomes? |
Getting Started
Start small. Pick one high-value business process where users are currently copy-pasting data from dashboards into spreadsheets. That's your first data application candidate.
The goal isn't to replace your BI platform — it's to build on top of it. The companies that figure this out first will have a significant competitive advantage. The ones that keep adding dashboards will keep wondering why their data team costs keep rising without corresponding business impact.