Guide to Building Effective Data Products

Organizations often struggle with poor user adoption of data and analytics solutions because they fail to consider the specific needs of data consumers in the design process. How many times have you built a fancy dashboard, only to have the data consumer ask, “where’s the button to download this to Excel?”

To ensure that you are answering the right questions and meeting the needs of your data consumers, consider their requirements before creating any data solution.

In this blog, we cover:

What is a Data Product?↵
How is a Data Product Different than Data as a Product?↵
How Can We Best Understand Data Consumers’ Needs?↵
What Should Be Included in the Packaging of a Data Product?↵
How Can We Start Creating and Using Data Products?↵
What are Examples of Data Product Use Cases?↵

What is a Data Product?

A data product is a cohesive packaged solution designed to meet the needs of its users. It is created with the intention of being reusable and having a positive impact on those who use it — and it needs to be managed after it is created to ensure its ongoing effectiveness.

A quality data product turns data into actionable information by being discoverable, interoperable, clearly owned, and continuously managed, while reducing the risk of errors and confusion caused by poorly designed data solutions not fit for purpose.

A data product can help to promote a culture of data-driven decision making across an organization because they are tailored specifically to their users.

Brown cardboard box opened with four white graphics inside and orange arrows coming out of the box pointing to the text: A data product includes example reports and dashboards, data freshness and location, field definitions, contact information.

A data product packages data along with the tools and resources necessary for the data consumer to effectively utilize it—ultimately allowing them to make informed decisions and drive business growth.

How is a Data Product Different than Data as a Product?

The key difference between a data product and Data as a Product is in their purpose and design.

A data product — which is a complete packaged solution—is created with the intention of solving a specific problem or meeting a particular need, going beyond just serving data as a raw asset. Data as a Product, on the other hand, is a subset of data products. Data products that are monetized or sold to external parties become Data as a Product — a third-party service offering of data.

An example of a data product could be a set of dashboards in an interactive mobile app designed internally for a team of real estate agents that contains all the information they need to negotiate offers. That same data product could be sold to external real estate agencies for use in their business — at which point it would become both Data as a Product and a data product.

Whether it sold to third parties or not, the key here is that a data product must be designed specifically with consumers’ needs in mind.

How Can We Best Understand Data Consumers’ Needs?

The first step to creating a successful data product is getting to know your data consumers and what information they need to be successful in their work. Here are a few key considerations that can help you gain insight into your data consumers’ needs:

Understand their use cases: Ask your data consumers about the specific tasks they perform with the data, with the intent to design a data product that is easy to use, automates their workload as much as possible, and provides relevant, useful, and timely information to them.
If all they do with a dashboard is export data for further manipulation, it’s clear that a dashboard isn’t the best way to serve that data to them. Ask questions like “Is your manual data manipulation something we could automate for you to make life easier and make the output more reliable?”If all they do with a dashboard is export data for further manipulation, it’s clear that a dashboard isn’t the best way to serve that data to them. Ask questions like “Is your manual data manipulation something we could automate for you to make life easier and make the output more reliable?”
Assess their technical abilities: Understand the technical expertise and data literacy level of your data consumers so that you can design a data product that is accessible to your audience, regardless of their technical background.
Don’t make people use a custom web-based API to access data if they are not technically strong. Find ways to allow them to access and manipulate data easily, provide resources to bring anyone up to speed, and continually check in to see where improvements can be made.
Learn about their decision-making process: Understand how your data consumers make decisions so that you can design a data product that is more likely to be adopted and used effectively by your target audience.
Don’t stop at the data consumers themselves — consider other teams downstream that may be impacted by the information gathered from the data product. When you understand the effect that information is having on other teams, you can proactively address needs throughout your organization.
Expect data products to be combined with one another. You must find ways to make your data product interoperable with downstream data. Data becomes more valuable when injected with business context, so combining data products is an activity that should be encouraged and expected.
Include the data, components, and instructions that other teams need to interface their data products with your data product. Remember that technical teams and less-than-technical business users may be consumers of your product, so the data product should be built to fit their needs and technical ability.

Four paragraphs alternating left to right vertically with snaked line between the paragraphs. Snaked line include navy blue, light blue, grey and dark navy sections. Above each paragraph states a way to understand data consumers’ needs to built effective data products.

Learning about your data consumers’ needs, pain points, and wants can aid you in creating effective data products.

What Should Be Included in the Packaging of a Data Product?

There are many ways to enhance the value of data by serving it as a data product to consumers. Consider including:

Information about the data itself — otherwise known as metadata — like how it was collected, who owns the process that created it, the source system it came from, how accurate or complete it is, and how often it will be updated.
Audience-appropriate documentation, such as business descriptions, field definitions, common calculations, common exclusions, common filters, and information about data lineage and provenance. Include how it was manipulated or augmented from its raw state
Tools and tips for how to best analyze the data, such as example reports, example dashboards, examples queries, code libraries, or example machine learning models to make the data product easier to use.
Optimal access methods to the data — sometimes this means an API or web-based interface, other times SQL-based database access is ideal. Other times it means just providing it as a spreadsheet in a file folder. One-size-fits-all solutions will not meet everyone’s needs in this area; it needs to be optimal to the data consumer.
Training and support through live sessions and recorded materials for data consumers to get answers to their questions. Identifying common questions is invaluable insight — they can be incorporated as consumer feedback to further improve and iterate on the data product.
Measurement metrics around adoption and use of the data product. Knowing who is using it is the best way to gather feedback on additional resources that could be helpful to them and other users.
Social communities around the data product. Create forums and platforms that allow consumers of the data product to help other users in a peer-to-peer social format compared to always needing to go through a central team for insights.
Clear communication of continual improvement and enhancements of the data product based on each improvement made by release over time.
Active security protocols, processes, and remediation measures to protect the data from unauthorized access, exposure, or use.

This is not an exhaustive list; there is so much more that can be included in a data product package — it ultimately depends on the best way to meet the needs of your data consumers.

How Can You Start Creating and Using Data Products?

Here are a few practical tips to creating and using effective data products:

Identify current pain points and needs of data consumers that data products can address. This could involve gathering input from stakeholders, analyzing data usage, or conducting market research. Ask questions like, “What is not working with the ways that data is delivered today?” “What is lacking in the current approach?”
Define the target audience so that you can tailor the data product to the needs of the end users and ensure that it delivers value. You cannot build something to delight an end user if you do not know who the end user is.
Build a cross-functional team because data products require input from a variety of departments and skillsets. This team should include data scientists, data engineers, dashboard developers, and subject matter experts.
Develop a plan for creating and launching the data product. This should include a communication plan to end users, not just technical artifacts. Yes, it is important to identify the data sources and tools that will be used, defining the scope of the product, and setting clear goals and milestones for project completion. It is much more important to understand how best to communicate with anyone that may be impacted by the information your data product serves to the organization during initial phases and ongoingly.
Test and iterate on the data product to ensure that it meets the needs of the end users. This always involves gathering feedback from stakeholders, analyzing usage data, and making refinements as needed. The first version of a data product will likely not be good enough—iterate quickly to build features that will add value and deprecate any that are not.

Talk to an expert about your data product needs.

Five light blue boxes laid out in a row of three and two, each states a step to creating and using effective data products. Each box has a dark navy icon.

Create effective data products by following 5 key steps that include knowing your audience and their needs and continual testing and iteration.

What are Examples of Data Product Use Cases?

Here are a few examples of data product use cases:

Customer churn. A software company can anticipate a customer’s cancellation of their service by looking at last login timestamps, number and type of support tickets, and social media data. A data product could be served to many internal teams like customer support, customer success, and account teams to proactively respond and reduce churn.
Consumer behavior. Customer data can be used to create segments to proactively recommend products or anticipate future purchases. For example, banks know if you have a mortgage with another company based on your account transactions. Providing customer segments as a data product could allow different divisions to use that data for their marketing and outreach purposes.
Employee retention. An HR-related data product can look holistically at employee data and flag anyone that is at risk of leaving the organization. Providing information on recent change in manager, last promotion, last pay increase, remote working situation, travel time to office from home address, social media, or market pay discrepancy could allow managers to proactively work to retain high performers likely to leave.
Sales performance. A data product that combines data on market share, sales leads, opportunities, and pipeline can help a business improve their sales performance. By using this data product, businesses can identify trends and areas for improvement, and take proactive measures to increase sales.
Marketing campaign optimization. A data product that analyzes data on website traffic, customer demographics, and customer behavior can help businesses optimize their marketing campaigns and measure their effectiveness. This information can then be used to make data-driven decisions, enabling businesses to improve their marketing strategies and boost their overall performance.

You Have a Data Product, Now What?

Once you’ve successfully released several data products across teams and business functions, you need to think very carefully about how data consumers can discover those products, combine them with other data products, and how to maintain standards of quality across disparate teams.

This is where strategies like data mesh or data fabric claim to address issues of scalability of data products in organizations. It is best to start small and get started with one or two data products before worrying about large-scale organizational strategy shifts though. And as with any data initiative, consider the core principles and practices that will enable successful outcomes.