This past week I put the finishing touches on my presentation for the 2014 Data Modeling Zone conference in Portland. My session talks about what the data architect can do to make their data models easier to understand. I am sharing my best practices on the use of color, text and graphics to deliver the data model’s story in the most appealing and effective manner.
A theme that echoes through this presentation is that the data architect needs to model to the audience. The style and content of a data model delivered to business users has unique needs that differ from those of the technical IT consumer. My data models are manifested in differing views specific to the target audience. Gone are the days when I delivered a single logical and physical model.
I am sharing six guidelines I follow when delivering data models to non IT clients. These are the basic building blocks I use to build data models that effectively reach business users.
- Easy on the eyes – The data model (ERD) should be readable and easy to digest. It is all about how to market the data model. Marketers work with plenty of white space with their message delivered in a pleasing pattern that catches the reader’s attention. This applies to how entities, relationships and modeling objects appear on the ERD.
- Talk business speak – Entities, attributes and relationships should be in words that are free from technical jargon, acronyms and abbreviations. They should also use the terminology, phrases and definitions that the business speaks. Writing in the business’ style conveys the story of the model with no translation needed.
- Right size the details – Model objects should show what the business user needs to know and not be overloaded with extraneous details. Modeling tools allow us to capture many characteristics of entities, attributes and relationships. Maintain the characteristics in your modeling tool. Display only the characteristics needed to convey your message.
- Group common things – Relationships become clearer when common objects are grouped as close to each other as possible. Don’t force the model consumer to follow many twisted and intersecting lines to make a connection. If the model becomes too complex, create subject areas to break down the complexity.
- Relationships that tell the story – Always include text on relationship lines. Relationships should form a sentence when paired with the parent and child entities. These are the business rules that bind the model together. A model with no relationship text cannot effectively tell the story.
- More than the model – Data models graphically organize data into structures and show the relationships that exist between these structures. It is a good snapshot of the data. Beyond that snapshot, it is the metadata behind the picture that tell the real story. Pay attention to the metadata collected and provide a means to present that metadata to the business in an easy to understand format such as a data dictionary or business glossary.
Tom Bilcze
I am presenting at Data Modeling Zone 2014 in Portland, Oregon. I hope that you will join me in my sessions: Is your data model a work of art? and Relationship versatility and the data modeler #DMzone