What are Data Products?

During the early 20th century, manufacturing automation provided organizations with the luxury of developing large suites of product offerings. With little thought to market-fit, consumers had a plethora of options to choose from. The large array of similar product offerings drove a highly competitive marketplace that in turn – fueled greater product diversity. However, in 1929, with the start of the Great Depression – if organizations wanted to survive, they needed to scale back their operations and be more deliberate with their product offerings.

It was during this time when modern Product Management fundamentals began to make there way into organizations. The role of the product manager was born out of the need to focus the manufacturing efforts of an organization to a suite of deliberate products. To do this, organizations made it the responsibility of Product Managers to understand a product through its entire life-cycle from conception, to design, marketing, and ultimately - market fit by tacking sales.

As the world began to discover digital technology, the role of the Product Manager continued to evolve. Branching out into more and more disciplines, product managers began incorporating new techniques to better understand consumer behavior. While these are foundational advancements in modern product management, they weren’t fundamental shifts in the way products were designed and managed.

Fast forwarding to today – modern products generate and consume more data than any point in time. This evolution in product design has led to a fundamental shift in product management. As products evolve with more personalized functionality, the need to focus and be more deliberate with product offerings has become paramount once again. Products that do not have personalized (data-driven) engines behind them will soon find themselves out paced and irrelevant.

We yet again find ourselves in a similar - evolve or go extinct type environment that spurred product management in the 20th century. However, rather than narrow the number of products offered, consumer demands now require a product to be unique and personalized to each individual customer. This shift in consumer expectations has dramatically changed the expectation of product manager. The need to be incorporate data as a core component of modern-day goods has led to the rise of the Data Products Manager.

With all new disciplines, the precise definition of Data Products has been quite fluid over the years. Having worked as a Data Product Manager and having reviewed hundreds of job postings for Data Products across organizations like Netflix, Bank of America, Walmart, and many other, the definition of Data Products can be organized into a few different categories – with one underlying theme – the use of data is one of the core functional components of the products ability to deliver value.

---

CONSUMABLE DATA PRODUCTS are the most common form that can be defined as, “A system that consumes data and applies a variety of transformations to deliver value.” You interact with these types of data products daily; for example, Amazon’s product recommendations, Outlook’s spam filter, Facebook’s post feed, Delta’s flight pricing, Google’s search and many more. These types of Data Products are so diverse that they require further classification.

Core: Products that contain minimal transformation of data to provide value; examples, Power BI dashboards, Google Analytics, Pivot Tables, and other forms of primarily data visualization and reporting.

Foundational: Products that contain mathematical transformation of data to provide value; for example, Discounted Cash Flow model for valuing a company, consolidated balance sheet, and business rule models - to name a few.

Growing: Products that contain statistical algorithms (machine learning) that consume data and produce data in a continuous feedback loop to deliver value; for example, Netflix’s movie recommendations, Google Maps route optimization, Visa’s fraud prevention, and Google Search. These products are learning and refining themselves from the action(s) performed using the products data output. It is through this feature in the data product that personalization to each end user can occur. For example, watching a certain genre of movie repeatedly and declining other genre recommendations will cause Netflix’s recommendation algorithm (Growing Data Product) to adapt and learn to recommend movies closer to what you have watched in the past.

Living: Products that a growing by design but also closely mimic narrow human intelligence. These data products interact and adapt with the physical world through quantitative and algorithmic sensing. Delta Airlines pricing optimization is an example of a living data product because it leverages reinforcement machine learning for dynamic pricing of it's airline tickets. This Living Data Product constantly adjusts prices for each customer segment based on a reward and penalty system that reinforces pricing that maximizes profits with respect too the current competitive environment (e.g. the pricing living data product is automatically adapt to meet a competitors promotion in a particular area without the need of a human to tell the algorithm a competitor is running a ticket sale).

PLATFORM DATA PRODUCTS are the tools most commonly used to develop consumable data products. They can be defined as, “Software enablement systems that consume and manipulate data through a suite of tools in order to deliver value.” If you are a data & analytics professional, you likely interact with at least one of these platform data products each day; for example, Power BI the tool is a enablement system that provides end users a suite of tools to manipulate data into a consumable data product. Other common examples can be found in all the cloud providers like AWS Lambdas, Microsoft Azure’s ML Studio, DataRobot, Cloudera Data Science Workbench, Databricks, Alteryx, and many more.

INTEGRATED DATA PRODUCTS are frequently living data products that are embedded into a physical device that interacts with an end user to provide value. Amazon's Alexa is perhaps the most well known integrated data product; with the physical device (speaker) functionally integrated with a living data product (Conversational AI) to create a symbiotic value. Other common examples include, Tesla's Auto-Pilot, Google Nest Thermostatic, Roomba, and my other "Smart" home devices.

It's clear, there has been yet again - a fundamental shift in the way products are designed and managed. To compete in this highly personalized environment - organizations must continue to design and develop all products as Data Products.

To help you get started or accelerate your journey, please reach out to the Data Products Group's at partners@dataproductsgroup.com .