Podcast: Play in new window | Download (Duration: 5:22 — 3.4MB) | Embed

Subscribe: RSS | More

Like so many of my generation, I watched in awe and fascination the historic Apollo lunar landing on July 20, 1969. This event paved the way for future space exploration, and the mission symbolized how anything is possible. Since then, we have launched probes to Mars and one to Jupiter. Nearly 30 years later, the Hubble Space Telescope provided a glimpse at what is probably a planet outside our solar system. In 2018, Astronomers discovered evidence for thousands of black holes located near the center of our Milky Way galaxy. Our view of the universe has significantly evolved since Copernicus (1500’s) successfully formulated a heliocentric model of our solar system. Every new piece of information builds upon the next, providing a more detailed view.

Astronomers built on this model, adding more and more detail. Today, if you look at the model of the universe, it is far more complex than the model first drawn by Copernicus. That’s how models work – as new data becomes available, models evolve to accommodate new insights. All of us in business and Marketing can look to the heavens to navigate how we develop data models. A model helps us organize and understand how different data elements relate to one another and the properties of the real world, and should be used to help make better business decisions.

We often talk with companies that are interested in creating data models. It’s common to hear the request for a model that looks a lot more like NASA’s than Copernicus’ model of the univers,e although many times, the data to create such a model does not yet exist. In situations such as these, we advise the following 5-step approach to the development of a data model.

Start with a clear understanding of what you want to model. For example, a model of the process customers go through to evaluate a solution or one for which of your customers are most likely to buy a new product, or used to identify new market opportunities or one that helps you model the contribution of Marketing to the company’s revenue.
You don’t have to start from scratch. You must start somewhere, but you don’t have to start from scratch. Conduct a literature review to find out what models are out there that might be like what you need, and consider using them as a base or starting point. If the model has been well vetted, this will help you gain internal buy-in and accelerate your efforts. Copernicus was able to leverage the work of others, including the work of Greek astronomer Aristarchus of Samos, who suggested 1800 years before Copernicus that the Earth and planets revolved around a stationary Sun.
Think like Copernicus. You’ll have to decide on your modeling methodology. Then create a high-level conceptual version of your model. Share it internally and make sure it captures what is intended. While the model is being reviewed, gain clarity around the data you will need to support the model. Remember to validate your data. Bad data will result in a bad model. Consider Plato and Aristotle’s first models of our universe, which placed Earth at the center and not the sun. The data wasn’t vali,d and as a result, neither was the model.
Start simple. Use the initial data to vet and employ the model. As you begin to use it, weave in your learnings and expand your level of detail. Built-in experiments that will help you gain more knowledg,e which you can then use to take the model to its next level. Make sure to gain buy-in at each stage.
Be willing to change. It’s an iterative process. You will learn as you go. It’s important to recognize that models change as information becomes more available. As NASA gains more information, it can fill in more details about space, stars, and planets. Maintain and scale your model to keep it relevant. This will most likely require you to invest in tools and processes.

Data modeling an essential key to success for supporting the business intelligence process. They are an integral part of capturing actionable intelligence to support effective decision-making and planning. Want to start building your data model?

FAQ:

(written by Penn of Sintra.ai)

Q1: Why are data models essential for business and Marketing decision-making?
A1: A data model helps you organize and understand how data elements relate to one another and to real-world conditions. The purpose is not complexity for its own sake; it is to improve decision-making by creating a coherent, testable representation of how something works.

Q2: What is the most important principle to remember about models?
A2: Models evolve. Just as our view of the universe expanded from Copernicus’ early heliocentric model to far more complex modern models, business models should expand as new data becomes available and new insights are learned.

Q3: Why do companies often ask for models that are too complex too soon?
A3: Because they want a “NASA-level” model—highly detailed and predictive—before they have the data, definitions, or operating discipline to support it. When the data does not yet exist (or is not valid), complexity creates false confidence rather than better decisions.

Q4: What is step one in building a useful data model?
A4: Start with a clear understanding of what you want to model—such as the customer evaluation process, likelihood to buy a new product, identification of market opportunities, or Marketing’s contribution to revenue. If you cannot define the purpose, you cannot define the data requirements.

Q5: Why shouldn’t you start from scratch when developing a model?
A5: Because vetted models already exist in academic and industry literature. A literature review can provide a credible starting point, accelerate progress, and improve internal buy-in—especially if the model has been tested and refined by others.

Q6: What does it mean to “think like Copernicus” when building a model?
A6: Choose a modeling methodology, create a high-level conceptual version first, and socialize it internally to confirm intent. In parallel, define the data required to support the model and validate that the data are accurate and fit for purpose—because bad data produces bad models.

Q7: Why should you start simple and iterate?
A7: Because early versions let you test usefulness with available data, learn what the model gets right or wrong, and add detail only when it improves decisions. Built-in experiments help you generate new knowledge, and staged buy-in reduces risk and accelerates adoption.

Q8: What does it mean to be willing to change the model over time?
A8: Treat modeling as an iterative, living process. As new information becomes available, update and scale the model to keep it relevant—often requiring investment in tools, processes, and governance so the model stays accurate and decision-grade.