The Data Modeling Paradox: Are we sacrificing insights for structure?

While companies heavily rely on data to inform their strategies, the very data models they employ often hinder their ability to ask deep and insightful questions.

The Data Modeling Paradox: Are we sacrificing insights for structure?

In today's data-driven world, it's no secret that companies rely heavily on data to make informed decisions. However, there's a growing concern that data models are actually hindering our ability to ask good questions. In this blog post, we'll explore the fascinating phenomenon of how traditional data models have unintentionally trained us to stick to simple and obvious questions, and why it's crucial to break free from this cycle.

So, let's dive in!

The Nuanced Curiosity of Newcomers

Have you ever noticed how fresh faces in a company, like the newly appointed VP of eCommerce, ask incredibly nuanced questions? They come armed with hypotheses, eager to explore every aspect of the data. They want to know about funnel conversions, user behavior, frequency of visits, and more. Their questions are rich in detail and demonstrate a deep understanding of the complexities involved.

The Data Modeling Paradox

Ironically, it’s the data structure itself that plays a significant role in suppressing our curiosity. When we pose nuanced questions to data, it becomes complicated. Bridging multiple systems or exploring nuanced aspects of the data takes time and often yields incomplete or inaccurate answers. Data teams struggle to handle these intricate inquiries, leading to longer turnaround times, multiple iterations, and unsatisfactory responses. Frustrated, we end up losing faith in our data teams and implicitly training our organizations to ask the "simple" data questions.

Tools and the Simplicity Trap

The tools we use to analyze and visualize data are designed primarily for creating dashboards and presenting simple, clean data. While they excel at providing straightforward and easily interpretable insights, they fall short when confronted with the complexities that arise from behavioral data and the need to integrate information from multiple systems. As a result, teams are often compelled to rely on basic, easy-to-answer questions that can be addressed within the limitations of these tools. As a result, the full potential of the data remains untapped, and more intricate and insightful questions remain unanswered.

The Role of Data Modeling

To break free from this cycle, we need a paradigm shift in data modeling. This is where the concept of an activity schema comes into play. The goal is to make complex questions as easy to answer as simple ones, enabling us to explore the full breadth and depth of our data. An activity schema seamlessly integrates data from various systems without the need for extensive data preparation, allowing for meaningful conversations around complex questions.

Imagine a new team member bursting with excitement, armed with endless nuanced questions they want to explore. With the right data modeling approach, they can start asking and answering those questions in just minutes, without feeling defeated by the need for simplicity. For instance, they can quickly delve into retention rates for the cohort of users who left 2-star support reviews, identify the top 5 referral sources that promote out-of-stock products, and test any hypotheses that come to mind, no matter how nuanced.


As the reliance on data models continues to grow, it's essential to challenge the limitations it inadvertently imposes on our ability to ask insightful questions. By adopting an activity schema-based approach and leveraging more flexible modeling techniques, we can unlock the true power of data.
So, let's embrace nuanced data questions and move away from the simplicity trap. By doing so, we open up a world of possibilities, enabling data-driven decision-making that truly drives innovation and growth.

Remember, the key to success lies not only in the data itself but also in the questions we ask and our ability to explore the intricate details hidden within.

Check us out on the Data Engineering Podcast

Find it on the podcast page or stream it below