Defining a Data Product

Defining a Data Product

Products can take many forms from the simple pencil, or a mobile application, to the complex airplane. The common thread throughout the entire spectrum of product complexity is the need for data products, the idea of a data product is the definition of value through data availability, enrichment, consumption and presentation. These data products become assets unto themselves to be traded or sold and often times will create value across multiple products, becoming a connective fabric for user experience & personalization.

Types of Products

As we think about how data products are defined, it is important to start with what other types of products are being defined and built. Each type of product has a specific set of characteristics of the personas that participate in the product creation and usage, the value measures of the product and the lifecycle of product engineering and adoption.

• Internal/Operational?- This is a category of product that are built to enable the back-office operations of an enterprise. These will often take the form of systems for managing corporate financials, internal collaboration tools, inventory management or customer relationship management. While these products often have small, relative user bases, they are critical to the successful operation of organizations.
• Data Product?- Our focus today, data products are sets of data that are ready for consumption by data consumers. They have a span of consumption methods including programatic, analytical modeling, visualization or raw human analysis. Data products act as connectivity between other consumer and user experiences.
• Mobile Application?- This is the product form that most consumers are familiar with, this category powers many interactions with value add for users. These mobile applications will both present data to provide engaging experiences while generating additional data about usage and personal preferences.
• Hardware Product?- This is a product domain that covers mobile phones, TVs and appliances. These products continue to be powered by data and utilize various connectivity options to provide advanced features for consumers. While the typical release cycle for a hardware product is slower than others, the ability to incrementally increase their capabilities through software advancements continues to accelerated, powered by data products.
• Analytical/Decisioning Model?- With the continued adoption of artificial intelligence across all types of products, these have become products themselves to be managed. They have a lifecycle, measures of success and managed experimentation to validate product assumptions. These models are the primary consumer of data products and drive much of the product definition for data features, velocity and quality.

This is certainly not a complete list of the types of products a product manager can imagine, but these categories will cover the most common products in the market today.

The most successful products will not fall into just a single category. Many product will span categories to maximize the number of target customers, or combine aspects of products to create a single experience that is highly valued and engaging. We define products in these categories as a way to think about how we divide up work between product teams, organize engineering specialties and define characteristics of specific products & sales strategies.

Data Product Dimensions

Data Products have several dimensions, and while many will overlap with other types of products, aspects like quality measures, time measures and business logic will be specific to data products in determining how to effectively define, test for acceptance by consumers and improve upon in future releases.

• Consumer Persona?- This is the primary user of the data product, they are the one that value is measured against. For programatic data products this could be an application or service, for data products that are analytical or operational in nature this will commonly be a role within the organization or customer base.
• Creator Persona?- The creator person is the individual or automated service responsible for sourcing the necessary data, executing the necessary transformations and producing an output that is utilized by the consumer persona. While automation is optimal for this aspect of product definition, manual creation is often acceptable during exploratory data analysis projects or experimentation.
• Owner Persona?- Many organizations have a data governance function to ensure compliance with legal requirements for data usage, privacy and disclosure. This persona represents the organizations data governance policies when data products are produced and consumed.?
• Quality Measures?- The level of quality, including completeness and accuracy will vary based on the consumers of a data product, and the target usage of decisions from the associated business processes. These quality measures have both a point in time component and a target for improvement over time. These are often key indicators of onboarding new users to an existing data product that require increased levels of data quality.
• Time Measures?- Many data products will have measures of time that determine if the data is usable for different problem spaces. These can measure both the time period that data is fresh enough for use, but also measures of velocity for data availability for consumption. Increased velocity of data becomes a key enabling for consumers to provide richer experiences using a data product features.
• Source Elements?- Source elements are the input data sets used to generate the data product. This list could be a single source requiring transformation and quality measurement and can grow in complex to multiple sources requiring integration, transformation and enrichment.
• Transformation Logic?- The primary value added by a data product is the transformation made to data, increasing its value for consumption. The data product must carefully document these transformations to ensure they are of high quality, understood by consumers and able to be changed rapidly for experimentation & implementation as business needs change the consumption of the data product.
• Exposed Features - The consumption of the data product will take place through the lens of features, these features can take on the form of specific data elements and an associated business definition. Features can be time-series data, elements of record or inferred elements. Each must be documented to identify the transformation logic used to create it and the time measures associated with the specific feature if different from the data product as a whole.
• Journey Touch Points?- Todays organizations are a complex set of interactions between personas, often referred too as journeys. These journeys are leveraged by product teams to understand how to enable more rapid adoption of capabilities and remove friction points from business processes & speed adoption. These journey touch points map to specific data products to identify their creation, enrichment and consumption.

Expanding on the consumer persona, who has specific needs as their role as the primary measure of adoption. The consumer persona becomes a key place for experimentation of new value in the data product and place for real usage feedback and feature requests.

• Business Process & Journey?- While the data product itself will have elements of intersection with various journeys in the organization, the consumer has very specific journeys they participate in. The definition of the data product will include these specific touch-points to ensure a measure of adoption is representative of the value of the data product.
• Business Decisioning?- The key reason for consuming the data product will be to make decisions by the consumer persona that affect the business in meaningful ways. The definition of the data product should capture these decisions to ensure the right data is available at the right time to enable the consumer persona to be effective at decisioning,
• Technologies of Comfort?- The level of technical comfort that a given consumer has will necessitate how the data is presented to the organization. Some organizations will prefer a visualization that is pre-built while others will demand raw data for analysis by different lenses and dimensions. The product owner for any given data product should focus on meeting users where they are today, eliminating the need for team members to develop new skills for the consumption of the data product.

Measuring Adoption

How that we have discussed what makes up a data product, how do we measure that we are integrating the right data, at the right time for the right consumers? The focus for measuring data products should be measures of adoption, and never measures of completion. With a proper data product, they are never complete, they are ever evolving for increasing their value and reach.

• Accessed?- How often is the data product accessed? This is a primary measure of how much value team members see from the data set and components. Measures of adoption over time show increasing value and can be used to asses if specific feature adds were successful and associated with increased usage.
• Consumer Driven Features?- Consumers of data products can span technical and non-technical teams, each will have a different perspective on their journey within the organization and how the data product is consumed for executing their portions of business processes. These consumers specifically asking for additional features is a measure of their engagement and desire to improve the tools around them.
• Velocity?- Many data products feed not just user experiences, but also analytical needs for model creation and training. The ability to increase velocity of data ensures up to date data for consumption and a richer experience for users that consume similar data from different systems.

Data Products are a linchpin, they span many other products including consumer facing applications, analytical modeling and production of physical products. Other measures of adoption can include measuring user experiences and engagement across different engagement points.

Do Data Products Have Roadmaps?

While I am not a huge fan of roadmaps and believe they are overused, there is a need in large, complex organizations to manage intersections between teams for releasing new capability and engaging in the organizational change components of product rollout. Roadmaps are a known tool for this activity to show aspirational goals over time.

If you do decide to create a roadmap for a data product, it should focus on elements that include additional data features, improvements to data quality and increases in data velocity. These are elements that will generate interest from additional consumers. Use roadmaps to show trajectory and direction, while avoiding specific commitments multiple quarters out so the engineering and product teams can maintain flexibility in responding to changing business conditions.

As organizations mature in their use of data, the definition and lifecycle management of data products ensures clear visibility to data available for decision making. The role of the product team is to capture the necessary elements of data products to show their input data, data transformation, consumers and associated journeys. Data products take both an internal and external view, treating them as a corporate asset with value to be increased over time, through effective feature enhancements, experimentation and investment.