Building your First Data Mesh Showcase
Data Mesh early adopters and “fast followers” have blazed ahead, taking some early big wins and enduring some pain along the way. We are now starting to see momentum across industries, in large and small organizations. These new adopters are partnering with experienced practitioners like Insight and exploiting the capabilities of innovative solutions like Starburst. If you are ready to get started, I recommend that you start by reading our Pathfinder eBook.
In this blog, we will cover some of the key topics we discussed in Episode 12 of Data Mesh TV, which you can watch here.
A favorite quote from the show, Ken said his biggest worry with Data Mesh is that “the business can’t consume the value fast enough.” Love it!
Selecting the First
Use Case Showcase
First, select a stakeholder with a mandate to succeed. You want to start with a small team that has an urgency to solve a business problem. Pick a stakeholder that will not allow you to fail. We discussed two different approaches and the pros and cons of each.
Starting with a Data Science team enables you to start quickly. They have the skills, they are likely already working within a domain model and they certainly have a backlog of data problems to solve. This team wants to work autonomously and they have the skills to exploit all of the self-service capabilities that you put in front of them.
On the other hand, that Data Science team may not possess the intimate business knowledge that is required to tackle some of the broader, strategic business objectives. Starting with a Business function allows you to create a more immediate business impact. These teams understand the business context for the data, and they have a keen understanding of what success looks like. They simply understand the problem with a business lens. If you can find a business team that is eager to work autonomously and willing to learn the skills to be self-sufficient, they would be an incredible partner.
There is no right answer, focus on the value. Don’t sell the value of Data Mesh promises, map the Data Mesh capabilities with teams that are asking and needing to work in this new way. The real magic happens when you can integrate the Business teams with the Data Science teams across closely integrated domains. They will have more success with becoming data-driven in the next two years than IT has done with data in the past ten years.
Start with an Innovation Lab
We also discussed the rise of building Labs or Innovation Domains, early in the journey. In a Data Mesh, you can design the governance and capabilities in this Domain to accelerate discovery and invention, free from the heavier governance or restrictions in other Domains. You allow these teams to move lightning fast, in an iterative manner. They can fail 100 times quickly, and discover their golden data product with incredible speed. If you want to build a competitive advantage with Data Products, focus on enabling teams to discover and build them with speed.
We’ve seen this model work, where the Innovation Domain hands over their “invention” to other domains where they go through proper governance and operational testing. The Innovation Domain in this case is only focused on discovery and invention, they are not impeded by the governance.
There are also Domain Labs that take on the full role of creating production-ready Data Products. This helps the Data Mesh drive greater consistency.
Don’t Change the Organization…yet
We both advise our customers that are early in the Data Mesh journey to limit organizational change. It’s hard to convince teams to change — it’s even harder when that change impacts organizational models. Our advice is to focus on a use case that is already working like a Domain. Sometimes these teams are referred to as tribes. The organizations that have built the fastest and most successful Data Mesh implementations, attribute their success to designing domains around existing organizational models. They may change later, but it is easier to optimize later when you are creating tangible value.
If you don’t know how many Domains you will need, focus on the one or three that can add immediate value today. Allow your Domain design to evolve organically, based on where value is going to be generated by Data Mesh ways of working. Some of the best Data Mesh designs have a handful of Domains working efficiently.
Data Mesh Pathfinder
Get Started On Your Data Mesh Journey