How to Create a Data Product

Creating a data product is complex endeavor involving many people and roles over multiple years, essentially the lifespan of a product. There are six main steps, ranging from defining the product, developing the product, packaging the product, governing the product, provisioning the product for consumption, and monitoring consumption and end of life. This is a virtuous cycle that loops back on itself. Product managers incorporate customer feedback and experiences to improve the product, and they develop training and support programs to ensure customer satisfaction and retention.  (See figure 1.) 

Product Packaging. There are three layers of metadata involved in the creation of a data product. Some metadata is attached to a data asset upon creation, some is added upon indexing by a data catalog, and the remaining is fixed when a data asset is published to a data store. (See figure 2.) 

Product Governance. Governing data products is a complex process involving multiple people, teams, and roles. A product team gathers requirements, writes a product plan, and markets internally and externally. Governance review boards evaluate whether the product is worth building as designed; technical review boards evaluate whether the product meets technical and design standards. Product teams incorporate customer and technical feedback into an ongoing planning and development process. (See figure 3.)

“A data product team is responsible for guiding a product from concept to retirement, including through multiple governance review boards.”

Figure 1. Steps Involved in Creating a Data Product


Figure 2. Layers of Metadata


Figure 3. Product Governance

View Large