How to Generate Business Value with Data Products

Data products are essential tools for companies looking to turn information into a competitive advantage. 

At Indicium, we believe that success in this journey relies on two complementary approaches: the Modern Data Stack (MDS), ensuring long-term technical sustainability, and data products, addressing real business challenges.

In this article, we explore how these strategies connect, what defines a data product, and how to adopt an approach that creates meaningful impact for your organization. 

Happy reading!

Data Solutions at Indicium

At Indicium, we believe there are two critical factors for the success of data solutions—those capable of delivering a competitive edge to companies:

• Successful data solutions are technically sustainable in the long term.
• Successful data solutions solve business problems.

With these two critical factors in mind, Indicium’s activities are guided by two key approaches:

• The Modern Data Stack (MDS), which focuses on technically sustainable solutions.
• The data products approach, which focuses on solving business challenges.

This article addresses some central questions about data products:

• What are data products?
• What types of data products exist?
• How can you adopt a data product strategy?

Let’s get started!

What are Data Products?

The concept of data products was first defined by DJ Patil, former chief data scientist for the United States, in his 2012 book Data Jujitsu: The Art of Turning Data into Product.

He conceptualized it as “a product that facilitates an end goal through the use of data”. 

The concept was also defined by Zhamak Dehghani in the article How to Move Beyond a Monolithic Data Lake to a Distributed Data Mesh, in which the author provides insights into how to create a decentralized data architecture to serve data to business teams: 

Domain data teams must apply product thinking [...] to the data sets they provide; considering their data assets as their products and the rest of the organization's data scientists, ML and data engineers as their customers.

Shall we think of an example?

Is Netflix a Data Product? 

No. 

The Netflix platform is not a product that facilitates an end goal through the use of data.

However, we can consider Netflix's content recommendation algorithm to be a data product.

This is because it helps the user choose a movie or series (end goal) through the use of data (history, degree of similarity, algorithm).

The authors define some principles that data products should follow.

• Discoverable: a data product should be easily discoverable, usually through a centralized catalog that lists all available data products with their information such as owners, origin, lineage, etc. By following this principle, it is possible to reduce data redundancy by ensuring that everyone in the organization is looking at the data in the same way.
• Addressable: a data product, once discovered, must have a unique address following a global convention to allow programmatic access. Organizations can adopt different naming conventions depending on the storage and format of the data. To facilitate use, it is necessary to develop common conventions in a decentralized architecture, where different domains can store and provide data sets in varying formats, with the aim of reducing friction in finding and accessing information.
• Trustworthy and truthful: there is nothing worse for a data consumer than not having data of the right quality for decision-making. Data products must be trustworthy and reflect the truth of the events or insights generated, with efforts focused on guaranteeing the quality of the data from its creation. 
• Secure and governed by a global access control: access to data sets must be secure, with access control policies applied individually to each data product. This principle guarantees the construction of safe-by-design data solutions and compliance with data legislation.
• Domain data cross-functional teams: teams that build data products need to be cross-functional, ensuring that there is a vision of generating value with data, represented by data product managers or data product owners, and a vision of product sustainability, represented by data engineering, analytics and AI positions. 

At Indicium, we have expanded the principles of data products.

In addition to the concepts in the literature, we have created and adopted additional principles to the data product methodology aimed at generating a competitive edge for our clients.

User-Centric Development

We believe that the development of data products should place data users and consumers at the heart of the discussion.

The concerns of data development teams should go beyond questions like:

• "Which tables should we integrate?"
• "How can we build incremental models more efficiently?"

Instead, they should focus on:

• "What business process generates this data?"
• "What challenges do decision-makers face related to this business process?"
• "Who are the users/personas that will interact with the data product, and what is their level of technical expertise in data?"
• "What are the key pain points and limitations in the current format of data consumption?"

We pride ourselves on having a multidisciplinary team of specialists and proprietary training programs that set us apart in the market.

Our expertise spans careers such as:

• Data Product Managers
• Data Experience Designers
• Data Consultants
• AI Engineers
• Data Engineers
• Analytics Engineers
• Data Analysts
• Data Scientists

Why Multidisciplinary Teams?

They deliver results and enhance organizational processes.

We believe data teams should support the business in decision-making to drive results. 

Beyond creating pipelines and data products, the data team is also responsible for integrating data product development into the daily operations of organizational processes.

After all, an attractive dashboard that doesn’t lead to any decision-making is a poor data product.

To achieve this, we use the Plan, Do, Check, Act/Adjust (PDCA) approach, where business goals are established, action plans with deadlines and responsibilities are created, and business indicators are monitored with relevant teams to ensure continuous improvement.

Data Product Adoption

Developing and maintaining data products is expensive.

It’s essential to monitor and manage data product adoption — tracking how many users engage, how often, and identifying the most and least accessed features.

In addition to monitoring, following the user-centric principle, continuous feedback is employed to evolve and improve data products, ensuring the right data is available in the right format for accurate decisions.

A Methodological Foundation

At Indicium, we’ve developed our own methodology for building data products: the 4D methodology.

We believe that a methodological foundation, honed through the experience of over 300 products for 60 clients, ensures the robustness and adaptability needed to create impactful data products.

What Are the Types of Data Products?

With the concept of data products and their principles in mind, they can be broadly categorized into five types:

• Raw Data: products that involve collecting raw data with minimal transformation or cleaning. These require users to put in the effort to generate value.
• Derived Data: products where data is transformed, cleaned, and enriched before being delivered to the user. Examples include star schema modeling or enriching operational data with customer segmentation models.
• Algorithms: data products that process data through computational algorithms and return new data or insights.
• Decision Support: products that help users make decisions without making the decisions for them, like dashboards and other visualization tools.
• Automated Decision-Making: systems that make decisions autonomously based on data, such as autonomous vehicles and drones.

Interfaces of Data Products

Beyond functional requirements, it’s important to categorize the interfaces that users interact with to generate value. 

Data products can have a single interface or a combination, tailored to technical or business users:

• Application Programming Interface (API): a technical interface for integrating data products with other systems.
• Visualization: dashboards and reports tailored to the technical knowledge of the user.
• Apps: applications that directly provide value to end users, simplifying the complexity of underlying data and algorithms.

Categories of Data Products

Combining product types with interfaces gives rise to various data product categories. Here are a few examples:

• A BI dashboard for visualizing company KPIs is a Decision Support data product with a Visualization interface.
• A data warehouse combines Raw Data, Derived Data, and Decision Support, with an API interface (e.g., SQL, ODBC).
• A business group model provided by big data companies is a blend of Raw Data Collection, Derived Data Construction, and proprietary Algorithms, typically accessed via API.
• A data app for collection planning leverages Algorithms through an App interface to facilitate better decisions.
• An autonomous car uses Automated Decision-Making with the App being the vehicle itself.

How to Adopt a Data Products Strategy?

According to Gartner, Data Products are one of the main technologies that are being adopted or will be adopted in a 12-month timeline by the world's leading companies. 

How Can We Adopt a Data Product Strategy and What Are Its Real Benefits?

At Indicium, we always look at organizational challenges from the perspective of People, Organization and Data (POD) - a methodology created by Indicium. 

Adopting a high-level data product strategy requires the following topics.

People

• Generate and disseminate knowledge among employees regarding technical knowledge related to data products.
• Create multidisciplinary teams that combine technical and business skills, introducing “non-traditional” data figures such as data product manager, data experience and data product analyst.
• Review data team organization models.

Organization

• Promote a data product culture, encouraging data reuse and collaboration between different areas of the company.
• Invest in leadership that supports and sponsors data product initiatives.
• Allocate a specific budget for the development and maintenance of data products.

Data

• Implement robust data governance processes, ensuring security, quality and compliance.
• Implement data ingestion, transformation, orchestration and cataloging processes with observability.
• Facilitate data consumption through intuitive, self-service tools and interfaces.

In other words…

An effective data product strategy creates real value by aligning data with business objectives.

Do you Want to Implement a Data Product Strategy?

Data products are essential tools for companies looking to turn information into competitive advantage.

From conception to implementation, Indicium transforms data into innovative products, tailor-made to meet the specific needs and challenges of each organization.

Do you want to unlock the true potential of your data and drive business growth? 

Our team is ready to create customized solutions that align technology, strategy and impact.

Click here to talk to us and find out how to turn data into real results.