Party at the Lakehouse: Building Better Data Ecosystems to Power Engagement and Decision-Making

A 2025 AIR Forum session titled “Party at the Lakehouse: Enhancing Data Ecosystems and Stakeholder Engagement” offered Institutional Researchers a compelling vision for the future of data architecture in higher education. Led by Henry Zheng, Matthew Hoolsema, and Roman Mitz (all from Carnegie Mellon University), the session used the inviting metaphor of a "lakehouse" to make a complex topic approachable, engaging, and practical. 

Combining the openness of a data lake with the structure of a warehouse, the “lakehouse” model represents an evolving strategy for managing and leveraging institutional data of different types and from a wide range of sources. But the session was about far more than technology. At its core, the conversation centered on how institutions can modernize their data environments to better support campus decision-making, democratize access, and foster a more responsive and collaborative data culture. 

“We’re not just here to talk infrastructure,” said Zheng. “We’re here to talk about impact - how to create a data ecosystem that is agile, inclusive, and built for action.” 

What Is a Lakehouse and Why Does It Matter? 

The lakehouse concept blends the best of two worlds: the flexibility and scalability of a data lake (which can handle a variety of data types) and the governance and performance of a traditional data warehouse. In the higher education context, it means giving stakeholders across the institution access to the data they need, when and how they need it, without compromising data integrity or control. 

This matters because the expectations placed on institutional data teams have dramatically increased. From enrollment forecasting to student success analytics, from employee retention to financial ratio analysis, from space utilization to funded research growth, decision-makers want actionable insights quickly, and they want to explore the data themselves. 

Matthew Hoolsema shared how Carnegie Mellon has reimagined its infrastructure to support multiple use cases without creating bottlenecks. “If you’re building a modern ecosystem, it has to work for everyone,” he explained. “That means IR, but also IT, faculty, advisors, deans, and students.” 

Data Culture Is the Foundation 

While the session addressed technical architecture, the presenters were clear: culture is just as critical as code. A well-designed data ecosystem is only effective if people trust it, understand it, and feel empowered to use it. 

Roman Mitz emphasized the need for IR and IT teams to work hand-in-hand, not just on infrastructure, but on training, governance, and communication. He noted that in many institutions, access remains siloed, data literacy is uneven, and analytics tools are underutilized. 

“You can build the most beautiful lakehouse in the world,” Roman said. “But if nobody knows how to get in, or is too intimidated to open the door, it won’t matter.” 

The CMU team shared that success depended on three cultural pillars: 

1. Transparency: Publishing data definitions, sharing decision rationales, and communicating clearly about what data is (and isn’t) saying. 
2. Inclusivity: Designing tools and platforms for a range of users, from advanced analysts to casual consumers.
3. Iterative Engagement: Continuously gathering feedback, adapting tools, and involving users in shaping the system.

Building a Better Data Ecosystem: A Practical Approach 

The presenters offered a roadmap for institutions seeking to evolve their data systems and culture. Their approach emphasized agility, adaptability, and above all, user-centered design. 

Key steps included: 

• Assess the Current Landscape 
  Map out your existing data environment—platforms, users, pain points, and aspirations. Know your starting point.
• Clarify Ownership and Access 
  Define who is responsible for what. A strong governance model prevents chaos without centralizing control.  Towards this end, CMU is developing a data analytics franchise model to address the ownership and access issues.
• Invest in Tools That Scale 
  Choose cloud-based platforms and data solutions that can evolve with institutional needs, supporting both structured and unstructured data.
• Design for Decision-Makers 
  Avoid overwhelming users with raw data. Provide dashboards, curated views, and tailored experiences aligned with specific roles and questions.
• Support Adoption
  Offer ongoing training, office hours, and embedded data support. Empower users to explore data with confidence.
• Start with High-Impact Areas 
  Demonstrate the value of the ecosystem by solving real problems—like improving retention, optimizing course offerings, or informing resource allocation.

Lessons from the Field: Data as a Strategic Asset 

The CMU team underscored that modernizing a data ecosystem is not a one-time project—it’s a long-term strategy that positions data as an institutional asset. Zheng encouraged attendees to think beyond compliance and reporting and begin framing data as an enabler of decision making and transformation. 

Examples included: 

• Embedding dashboards into daily workflows for deans and department heads 
• Using predictive analytics to target interventions with at-risk students
• Sharing exploratory data tools that allow users to answer their own “what if” questions
• Making all major enterprise data assets (HR, finance, research, space, etc.) accessible from a single reporting portal.

Perhaps most importantly, the presenters emphasized that success requires cross-functional collaboration. IR teams alone cannot transform data culture, but they can lead the way. 

“The lakehouse isn’t just a structure,” Hoolsema said. “It’s a mindset—a shift from gatekeeping to enabling, from reports to exploration, from silos to shared success.” 

Final Thoughts: Everyone’s Invited 

The session’s metaphor—throwing a party at the lakehouse—proved both apt and memorable. Creating a welcoming, functional data ecosystem isn’t about perfection. It’s about creating a space where people want to be, feel included, and know how to contribute. 

Whether you’re at the beginning of your data modernization journey or refining an existing system, this session offered both strategic vision and tactical advice for institutional researchers. The message was clear: a well-designed lakehouse can support a thriving, dynamic data culture—and everyone’s invited. 

Note: Contributing author to the AIR Forum impact session: Alexis Parker (Carnegie Mellon University) 

2026 AIR Forum

Mark your calendars to join us May 26–29 in Washington, D.C., for the 2026 AIR Forum! Pre-conference educational opportunities will be offered May 25–26 for an additional fee. 

With more than 200 crowd-sourced sessions by colleagues and thought leaders from around the globe, AIR Forum is the “must-attend” event for higher education professionals who build and support data-informed decision cultures.