Why data products matter

The increasing interest in data mesh architectures has driven more and more conversation around the concept of data as a product. Yet this idea didn’t originate with data mesh. Long before data mesh gained traction, data professionals considered applying product thinking to their distributed datasets. 

The core appeal of transforming datasets into products that can be consumed in a simple, secure way inside or even outside the organization is obvious (i.e. the opportunity and development of new products, services, breathe legacy business new life, and so much more). What the rise of data mesh has done is give this data-as-a-product approach more momentum. 

At AWS re:Invent 2022, theCUBE industry analysts Dave Vellante and Lisa Martin sat down with our CEO, Justin Borgman, and Ashwin Patil, Principal, AI & Data Engineering at Deloitte, to discuss how organizations can take advantage of data products — and why they should do so in the first place. We encourage you to watch the full interview here, but for this post, we’ve distilled it to a few key takeaways. 

Why you need to productize data

The term data products might suggest this approach is focused entirely on monetizing your internal datasets by making them available for external consumption. That’s certainly one aspect, but as the experts point out at the beginning of the interview, there’s tremendous value internally as well. 

On one level, transforming your distributed datasets into data products is really about facilitating self-service consumption, or making it easier for different people within your organization to unlock hidden value or insights in all of your data — not just the datasets close to them or their line of business. 

“Data within the enterprise can sometimes be fragmented. And data products bring all this data together,” Ashwin notes, “and this drives better, faster decision making.”

Data products serve as a bridge between data mesh and data fabric

The differences between data mesh and data fabrics continue to confuse and Vellante circles back to why these concepts are taking hold today, and how data products fit. 

Justin explains how there has been a disconnect in organizations between the people who know and manage data infrastructure and the analysts who want to access that data to answer questions. Data mesh architectures and the creation of curated data products within a mesh both function as a bridge that brings these two groups closer together. They provide a framework for thinking about and extracting insights from all of your datasets.

Interestingly, Ashwin later points out that he sees data products acting as a bridge between the data fabric concept and data mesh architectures. “Data mesh creates a structure that works complementary to the data fabric that then lends itself to data products,” Ashwin says, “So data products become a medium which drives connections between a data mesh and data fabric into the real world for usage and consumption.”

Getting started quickly: productizing your data 

The interview hits on a number of different topics, including the role of the data analyst in this evolving space, the increase in citizen data science, and the importance of governance as data products become democratized. 

Ashwin talks about how he expects to see more companies going beyond their four walls to find more usable, relevant data to help them make better decisions. Whether you’re talking about data fabrics, data mesh, or data products, the transition isn’t necessarily any easy one. Technical, organizational, and operational changes may be required. 

Yet the advantages are clear, so theCUBE’s Lisa Martin brings the conversation back to a practical level near the end of the interview. What can companies do to productize data and unlock business value? 

In response, Justin offers a concise summary of what companies can do today to start productizing their data so that they can unlock new revenue streams and new routes to market. First, find data that is valuable to the consumer within the business and create a product with it. Create a view of your data that can span multiple data sources; you might take a table from a traditional data warehouse, a table from an operational system (ie.MongoDB), and a table from a data lake — and they can all join together to build one, easily consumable view. 

As Justin points out, the shift to data mesh and data products is a journey, not a quick transition, and Starburst offers organizations an easy, but powerful way to get started.  

Learn more about data products

There are more great insights in the interview and we will be going into greater detail on all these topics at Datanova 2023. If you haven’t registered yet, sign up today!