Chapter 1 (The Era of the Mixed Model Artist) draft is done, and I’m waiting until tomorrow to publish it because there might be a few things to tweak.
Here’s the part where I define Mixed Model Arts, the core of this book series.
Chapter 1 will be available to paid subscribers. If you haven’t become a paid subscriber, please do so. You’ll get access to the new revised chapters coming out (there’s a ton coming this month and next), an electronic copy of the book when it’s released, and access to data modeling digital courses.
Thanks,
Joe
Modern data modelers need to live at the intersection of business and tech. Ellie.ai allows you to collaborate effectively with business while maintaining credibility with the Tech team. Get contextual support from AI, reverse engineer anything building a repository of sources with synthetic AI generated contextual metadata while delivering insights via an MCP Server and integrating anything with full blown API support.
The full-stack data modeling future is here today!
Thanks to Ellie.ai for partnering with Practical Data Modeling.
The best fighter is not a Boxer, Karate or Judo man. The best fighter is someone who can adapt on any style. He kicks too good for a Boxer, throws too good for a Karate man, and punches too good for a Judo man. - Bruce Lee
Bruce Lee was the first Mixed Martial Artist, studying various martial arts and combining them into Jeet Kune Do in the 1960s. Unfortunately, it took decades and the UFC for his philosophy to become mainstream. In combat sports, Mixed Martial Arts emerged when fighters realized that no single discipline was sufficient for the complexity of a real fight. In data, Mixed Model Arts emerges from the same recognition: no single modeling approach is sufficient for the complexity of a modern data ecosystem.
Mixed Model Arts is a way of data modeling that combines techniques from many areas—structured, semi-structured, and unstructured; transactional and analytical; for both people and machines—to create coherent data models that meet all the needs of today’s data world.
Let me unpack this definition, because every word matters.
Combines techniques from many areas. A Mixed Model Artist doesn’t ignore any tradition. They learn relational, dimensional, document, graph, feature engineering, and semantic modeling, plus anything else that helps in a given situation. They know the strengths and weaknesses of each. Instead of seeing these as competing beliefs, they treat them as tools in one toolkit. Like an MMA fighter who learns boxing, wrestling, and jiu-jitsu for different skills, the Mixed Model Artist studies each modeling style for what it does best.
Structured, semi-structured, and unstructured. Real data comes in many forms. Tables, JSON documents, text, images, audio, video, and embeddings—a Mixed Model Artist can model across all of them. This doesn’t mean every modeler needs deep expertise in every form. But they need sufficient cross-form literacy to make informed decisions about how different data forms relate to one another and to the business reality they collectively represent.
Transactional and analytical. The same business reality—a customer, an order, a product—is modeled differently depending on whether you’re focused on writing data (transactional) or reading data (analytical). A Mixed Model Artist understands both approaches and can connect them. They don’t dismiss the warehouse modeler’s star schema or the application developer’s relational model. They know why each one matters.
For both people and machines. This is the new frontier, and it’s what makes this moment in data modeling history unique. Data models now serve both human analysts (through dashboards, reports, and ad hoc queries) and AI systems (through semantic layers, feature stores, and knowledge graphs). The mixed-model artist builds models that work for both audiences, ensuring that the semantics a human analyst understands are the same as those an LLM uses to generate a SQL query.
Coherent data models. This is the most important part of the definition. The goal isn’t to have separate, disconnected models for different needs. It’s to have models that fit together, share the same definitions, refer to the same entities, and agree on what terms like “customer,” “revenue,” and “order” mean. Coherence is what sets Mixed Model Arts apart from chaos. It doesn’t mean building one giant, monolithic model that rules them all. It’s what makes a mixed martial artist different from someone who just swings wildly.
If you’ve been paying attention, you’ll see the main theme is: It Depends. The Mixed Model Artist doesn’t have one answer for every problem. Instead, they use a framework to find the right answer based on the situation, constraints, and the intended users of the model. The difference between a one-dimensional modeler and a Mixed Model Artist isn’t about having opinions—it’s about not being held hostage by them. It’s about knowing what matters in each case.
Practical Data Modeling is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.