Data Products Part II: Data Products Require Product Thinking
ABSTRACT: The hardest part about implementing data products is fostering a product mindset among the people responsible for defining, governing, building, and shipping data products.
According to a recent LinkedIn poll I conducted, 54% of respondents said their organization has implemented data products. That’s a surprisingly high percentage, and it led me to ask how people define a data product. As I said in my prior blog, a data product is not a data asset with better quality or governance. It’s the output of a data product organization that oversees the complete lifecycle of a data product and manages its availability in an automated way.
The organizations that have implemented bonafide data products say that the biggest challenge is creating a “product mindset”. This is ironic because all commercial organizations have product teams that define and package products and services. But product thinking and experience are almost non-existent among data and analytics professionals. Hence, adopting a data product strategy is an uphill climb for most data teams.
Adopting a data product strategy is an uphill climb for most data teams.
So, how do you instill a product mindset among data people? The chief data officer can’t just declare that the data team is now a product organization. They can’t simply appoint people to serve as product managers and owners and expect that a product culture will grow in alien soil. Without ample education and coaching, data professionals and the businesspeople they support will simply resort to old ways of doing things. If there is a product team, it will be in name only.
Here are a few tips to nurture product mindset and develop a bonafide product organization:
Seed product experts on teams. The best way to create a product culture is to hire or assign people who already have a product mindset and know what product teams do and don’t do. Of course, this can be prohibitively expensive if you have more than a handful of domain teams interested in producing data products. To reduce costs, you could steal product managers from elsewhere in your company, but you won’t make many allies this way.
Train and coach. You can also hire a product expert to train and coach newly appointed product managers. The coaching element is critical since new product managers will need guidance long after they’ve completed their training. To maintain a product sensibility, they will need constant reminders and hints to keep from falling back on old habits.
Centralize. Some organizations create a central product team that fields requests from business domains, creates product plans, and oversees the execution of the data products. A centralized approach addresses the resource constraint, but it creates a potential bottleneck that could hamper the development and delivery of data products at a pace that meets organizational needs.
To instill a product mindset, you should seed product experts on teams, train and coach new product managers, and create a central product team.
The good news is that creating a product organization is not rocket science. It doesn’t take an inordinate amount of time to train and coach product managers. However, it’s imperative that the organization properly defines roles and processes to support the management and marketing of data products.
The first step in becoming a data product organization is to identify the roles required to create, govern, and manage data products. The next step is to assign the roles to individuals. In small or nascent environments, a single individual can serve most, if not all, roles. In larger companies, there might be two or three people per role.
The most critical roles are the product owner, product manager, product marketing manager, and product developer.
Product Owner. This person is accountable for setting product goals and delivering results. Usually a high-ranking executive, a product owner:
Evangelizes the need for the data product.
Secures sustained funding.
Oversees the product team.
Updates the executive team.
Product Manager. This person is responsible for developing and enhancing the data product and does the following:
Gathers requirements from potential customers.
Defines product scope, features, and roadmap.
Coordinates development with a technical team.
Shepherds the product through the governance process.
Monitors product issues and fixes.
Consolidates feedback from customers.
Maintains a roadmap of enhancements.
Product Marketing Manager. This role can be merged into the Product Manager’s job, especially in small companies. The product marketing manager:
Publishes the data product to a data product platform.
Promotes the data product through appropriate internal or external channels.
Monitors product usage.
Solicits feedback from customers and passes on to the product manager.
Works with “sales” to drive usage or subscriptions.
Product Developer. A product developer is usually part of the data or IT team. The developer leads a team of people with various skills, depending on the nature of the data product. This team:
Translates business requirements into technical specifications.
Sources, models, and transforms data.
Builds data pipelines and data flows.
Builds analytic applications, e.g., dashboards and data-driven solutions
Product Review Board. The product review board is a cross-functional team of business leaders who review product proposals to ensure they align with organizational goals and objectives, offer a measurable return on investment, and deliver sufficient business value. This team will accept or reject proposals or ask the proposing team to modify their application to better meet the board’s criteria. The board may also have a fast-track process for specific types of product proposals, such as enhancements to existing products or data products shared by just two or three domains.
Technical Review Board. This review board consists of technical leaders who review built products to ensure they meet business and technical guidelines. This board will evaluate a product’s performance and scalability, metric calculations, graphical design, data sources and transformations, access and security controls, and data platform and tools. It will assess whether the product can run as is or needs to be migrated to a more standard enterprise data platform and whether the data exists in the enterprise architecture or needs to be sourced anew.
The hardest part about implementing data products is fostering a product mindset among the people responsible for defining, governing, building, and shipping data products. To create a product culture, an organization needs to either seed its initial product teams with seasoned product managers, provide substantial training and coaching, or centralize product management.
It’s also important that an organization create processes to facilitate the work of the product owner, product manager, and product marketing manager as well as the technical team that builds and fixes the data product and governance boards that review product compliance. We will describe these processes in the next article in this series.