Data Usability: How to Build Better Data Products?

Looking ahead to the future, the utilization of data is poised to become a prevailing norm, empowering intelligent workflows and effortless human-machine interactions. McKinsey’s projection posits that by 2025, the prevalence of these seamless interactions will rival that of a corporate balance sheet, ushering in heightened productivity.

In materializing this vision, a crucial aspect is the consideration of data usability. Data products should not merely consist of reusable datasets tailored for specific objectives; they should also be easily consumable through tools that are intuitive and comprehensible within the end-user’s working context.

This article initiates by introducing a ‘data to impact framework,’ which illustrates how humans leverage data products — originating from AI and data pipelines — to transform raw data into tangible outcomes. Subsequently, data usability emerges as a pivotal factor in the construction of better data products. By incorporating data usability within your organization’s data strategy, you can initiate the journey towards augmenting the impact derived from data-driven practices.

Data usability will facilitate humans to work better with data products, created by automated pipelines and AI, within their context of use. In time, this will help to get more impact from data [Image: Author].

Data To Impact Framework

Organizations strive to leverage data effectively to achieve tangible impact and value. This impact, which can take various forms such as increased revenue, risk reduction, or improved quality, plays a crucial role in shaping the success of any venture. Data can also contribute to making a company more appealing through personalized marketing or offer motivation and hope, as seen in cases like Strava’s condition metrics or advancements in proactive cancer screening.

To realize impact from data, three key components come into play:

1. Technology: Automated pipelines extract, transform, and load data as valuable information into data consuming applications. AI Algorithms can then further process data from these pipelines, turning it into more actionable insights. For example, a pipeline extracts data from a CRM and web tracking system, which is further processed by an AI algorithm in a dataset identifying potential churners.
2. Data Products: Reusable data assets tailored to deliver reliable datasets for specific purposes. Pipelines and AI algorithms produce data products, ensure its quality, and make it accessible and comprehensible to users with distinct needs. Typically centered around specific business entities like customers or orders, data products encompass metadata and dataset instances.
3. Humans: Human analysts play a pivotal role in making the right decisions based on the data products created by the automated pipelines and AI algorithms. Ultimately, these decisions drive impact for the organization.

The data to impact framework shows how automated pipelines, AI and humans work together while turning data — by means of data products — into impact [Image: Author].

The data to impact framework illustrates how these three actors collaborate to turn raw data into tangible results. Initially, automated pipelines process and prepare data products. Next, human analysts analyze this data and make informed decisions that lead to impactful actions. AI and automated pipelines can assist humans seamlessly in optimizing the decision-making process. In some cases, the entire data-to-action process can be fully automated by pipelines and AI without any human intervention.

Let’s explore four examples, also depicted on the figure below, to illustrate our data to impact framework:

1. Raw Spreadsheet Analysis: Humans manually export data from applications, like a CRM system, into spreadsheets, where they cleanse and analyze the data to gain insights. These insights inform the actions that lead to impact.
2. Reporting Tools: Data is automatically preprocessed at regular intervals and loaded into reporting tools. Users can utilize predefined dashboards and interactions to make well-informed decisions.
3. Digital Applications: Customer service employees receive AI-generated indicators (e.g., green or red lights) in their service desk applications, predicting whether a customer might churn. Based on this information and the context of a customer call, the employee decides whether to offer a discount, impacting the overall customer base.
4. Autonomous Decisions: AI algorithms continuously monitor customer behavior to predict potential churn. When certain parameters are met, the algorithm automatically triggers an email campaign with discount proposals, took prevent the customer to churn.

Four examples to illustrate the data to impact framework: (1) raw spreadsheet analysis, (2) reporting tools, (3) interactive tools and (4) autonomous decisions [Image: Author].

Data Usability

Enhancing data usability results in the creation of more impactful data products, fostering a ripple effect that cultivates intelligent workflows and seamless interactions between humans and machines. Data usability extends from the broader concept of usability, which is defined by the Interaction Design Foundation as:

“A measure of how well a specific user in a specific context can use a product/design to achieve a defined goal effectively, efficiently, and satisfactorily.”

Based on this general definition, we can infer data usability as:

“A measure of how well a user in a specific context can use data to achieve a defined goal effectively, efficiently, and satisfactorily.”

Within the context of our data-to-impact framework, several elements play a vital role in enhancing data usability:

1. User — Understanding the end-user is paramount. The better we know the individuals we are building data products for, the easier it becomes to make those products more user-friendly. Elements 2 and 3 delve deeper into understanding the user.
2. Persona — Personas describe the types of users we cater to in the data world. These personas encompass characteristics such as attention to detail, data literacy levels, and whether data is used for individual research or team collaboration.
3. Process & Context — It is crucial to consider the processes and contexts in which end-users work. For example, some may require quick answers to specific data queries in operational processes, while others may be exploring data to uncover trends and patterns. Understanding the functional domain within which users operate is also essential.
4. Impact — Determining the user’s data goals and what they aim to achieve is vital. Identifying the role of data in helping users reach their objectives is key to optimizing data usability.
5. Data Product— It is essential that data products shape data in a way that is easy to comprehend for each end-user. Mapping data to known domain concepts makes it more accessible for end-users.
6. Tool — Selecting the right tool for users to access and work with data efficiently is critical. The term “tool” encompasses various technological components, including data transformation pipelines, artificial intelligence modules, and user interfaces facilitating data interaction.

Data usability can be increased by tuning every element that contributes to the way people interact with data [Image: Author].

Building Better Data Products

Data usability plays a crucial role in making data products more user-friendly and easier to comprehend within their intended context of use. This involves a thorough assessment of the personas who will utilize the product, the context in which they operate, and the goals they seek to achieve. Understanding these elements enables data products to be tailored to users’ needs and seamlessly integrated into tools that enhance their usability within their specific context.

Focusing on Data usability helps to build better, more impactful, data products [Image: Author].

Let’s explore some examples that illustrate how specific tools foster data usability:

1. Environmental Social Governance (ESG) Data Lab — As more companies focus on ESG parameters, data scientists often have to analyze various ESG-related data sources to gain insights and detect trends. A valuable data product for them would be ESG-related data assets readily available in their workbench, such as notebooks or containerized setups. This streamlined access facilitates data discovery and empowers data scientists to excel in their tasks.
2. KPI Reports — Data products containing Key Performance Indicators (KPIs) are often consumed in reporting applications. These applications enable managers to closely monitor KPIs and make data-driven decisions at the right time.
3. Citizen Data Engineering — In the Data Mesh paradigm, business teams handle some of their own data engineering tasks, like loading and transforming data. For better usability, data assets should be made accessible in user-friendly toolsets, allowing business users to perform these tasks effortlessly. For instance, a visual pipeline editing tool can be more approachable than a code-based solution.
4. Digital Application for Tumor Detection—Radiologists use digital applications to detect breast cancer in mammograms. AI algorithms can conduct an initial analysis of the raw images and highlight potential malignant areas in the scan. This integration of data products produced by AI algorithms with digital radiologist applications enhances the efficiency and accuracy of tumor detection.

Examples of how data products are being consumed by different types of users in different technologies and contexts of use [Image: Author].

Data Strategy

The pursuit of usable data products demands a collaborative effort beyond the confines of conventional data teams. The achievement of enhanced data usability hinges on bridging the gaps between business, IT, and data teams, fostering an environment where data products crafted by one team seamlessly align with tools developed or maintained by others.

To break departmental silos and cultivate a culture of cohesion, the creation of a widely embraced data strategy becomes paramount. Several illustrative approaches can guide the development of such a comprehensive data strategy:

1. Adaptive Budget Allocation — Elevating data usability necessitates ensuring that data products are accessible through tools tailored to specific user needs. This collaborative endeavor calls for a budgeting approach that offers financial flexibility, enabling multiple teams to draw from the available resources. Overcoming the limitations of fixed yearly budgets that hinder resource sharing is pivotal for enabling effective cross-team collaborations.
2. Embracing Enterprise Architecture (EA) Principles — Adopting EA principles can serve as a strategic compass for evaluating projects and software products in terms of their potential contributions to data usability. For instance, tools equipped with robust APIs for seamless data integration might be favored over alternatives lacking the means to effectively incorporate external data.
3. Empowered Data Product Ownership — While data product owners traditionally focus on delivering reusable data assets within their respective domains, achieving optimal data usability requires extending their purview. This broader responsibility entails accounting for diverse data product consumption patterns. Empowering data product owners with ownership across teams enables them to surmount barriers, proactively address impediments, and foster seamless collaboration to realize their objectives.

Conclusion

The journey from data to impact is multifaceted. It involves the orchestrated collaboration of technology, including automated pipelines and AI, to produce data products that are then leveraged by humans. Data usability plays a pivotal role in simplifying the consumption of these data products, ultimately enhancing their potential impact.

I strongly recommend incorporating data usability into the organizational data strategy. This strategic integration can effectively break down departmental barriers and foster a cohesive culture.

Questions? Feedback? Connect with me on LinkedIn or contact me directly at Jan@Sievax.be!

This article is proudly brought to you by Sievax, the consulting firm dedicated to guiding you towards data excellence. Interested in learning more? Visit our website! We offer a Data Strategy Masterclass that provides a deeper understanding of the world of data strategy.