Data science and user experience design have recently become critical to the success of companies' sophisticated internal digital tools for analyzing, processing information, and predicting what-if scenarios. Used by internal teams and operations staff, these tools aim to optimize processes, predict outcomes, and inform decision-making in increasingly complex systems. For those working in data science, a common approach is to view humans as “human-in-the-loop,” meaning users provide inputs to a system or validate outputs, but their role remains secondary to technology and algorithms. It is worth considering that the complexity and nuanced requirements of modeling tools have begun to require a shift to a more human-centered approach to make the tools accessible to a wider audience. This article examines human-centered design and the need for such a focus for those working on data science projects and suggests strategies to facilitate this transition. Determining the Difference To address human-related issues (emotional state, learning, outcome), the concept of “human-in-the-loop” (HITL) was introduced, which involves incorporating human knowledge into the modeling process. The HITL approach is often a common way of thinking among data science teams, where humans are seen as secondary players who help the machine function better or validate the results generated by algorithms. This interdisciplinary approach crosses computer science, cognitive science, and psychology and consists of three components: data preprocessing, data modeling, and process modification to improve performance. However, it is important to consider that the results and performance of ML models can be unpredictable, so it remains unclear which aspect of human-machine interaction yields the best learning results. People are thus viewed as add-ons to the system—sources of labeled data, corrections, or ideas that help fine-tune the models—without taking into account their broader needs, motivations, or limitations. Human involvement depends on the type of solution being implemented: strategic, tactical, or operational. In operational and tactical systems, the human role is usually minimal. In contrast, in strategic solutions, the human role is critical because these decisions require context, explainability, and domain expertise. They often rely on human interpretation to validate assumptions, assess risks, and align outcomes with broader goals. Human-centered design, on the other hand, places the user at the center of the design and development process by aligning technology with human needs, behaviors, motivation, and goals. Understanding the "why" behind user behaviors is a key component of human-centered design, which integrates user context and requirements thoroughly into all phases of product development. For data scientists, embracing human-centered design is more than just incorporating human input into systems; it also entails actively reshaping systems to meet the actual users’ needs and developing tools that empower rather than burden them. Why Human-Centered Design Matters In complex, high-stakes environments, inconsistent forecasts, confusing workflows, or overly complex interfaces can lead to poor decision-making, frustration, and operational inefficiencies. That’s why a human-centered approach is invaluable in such settings, enabling users to receive answers tailored to their needs. Often, users view data science solutions as “black boxes,” making it difficult to trust them or understand the results they produce. Developing tools and solutions that explain the relationships between inputs and outputs and the behavior of models can improve trust and usability, empowering users to confidently make informed decisions. This need for explainability and transparency is directly related to the need for human-centered design and more understandable solutions. When tools are designed with end-users in mind, they are more likely to be adopted and used effectively by users. When data scientists understand the nuances of user tasks, they can create models and interfaces that match users’ real workflows and contexts, rather than idealized or abstract ones. This increases trust in the tool’s output and minimizes the learning curve for users. Moreover, the accuracy and applicability of machine learning models can be highly dependent on the user context. These subtleties are better captured by a human-centered approach, which ultimately produces more accurate and meaningful predictions for users. By providing users with the tools they need to make informed decisions, human-centered design empowers users. Instead of relying on the user to validate a machine’s predictions, tools can allow users to make decisions alongside the machine, fostering collaboration rather than a directive relationship. Practical Steps to Move Toward Human-Centered Design While the benefits of human-centered design may be clear and obvious, shifting from a “human in the loop” requires both structural changes and cultural and team support. Here are some actionable steps teams can take to embrace human-centered design. 1. User Research First Researching users and their context is fundamental to human-centered design. Thus, interviews, ethnographic studies, observations of users’ workflows, and identification of pain points, motivations, and goals help to better understand and study users. Such research produces qualitative insights that can be complemented by quantitative data, providing a deeper understanding of users' context. Additionally, creating user journey maps that detail the workflows and touchpoints users have with various tools and other organizational functions allows for a more accurate assessment of what users are experiencing. This visual tool can help highlight opportunities where these tools can add more value or streamline user processes. 2. Prototype With a Purpose To make efficient use of a team's time by experimenting with different models, assumptions, and data visualizations while focusing not only on achieving technical goals but also on creating features that are clear, relevant, and useful to the user, rapid prototyping can help. It’s a fundamental practice of human-centered design that brings ideas to life and allows users to test the prototype early, for example in Figma, without spending time on development. This ensures that user feedback is obtained before a huge amount of time is spent on product development and helps to quickly get to the point where the user's problem is solved by matching the real needs and defining what tasks the user needs to perform to achieve it. 3. Design for Trust and Transparency A critical component of human-centered design is building trust and transparency so that when users receive forecasts or recommendations, they understand the validity of those results and can make business decisions and act on them with confidence. Trust in data tools can be achieved by helping users understand why the system suggests a certain outcome or path and by visualizing results that are understandable to the user and consistent with their mental model. 4. Development of a Collaborative and Interdisciplinary Team Culture Creating successful human-centered data tools requires collaboration across disciplines. Having a shared team vision based on the human-centered design will help bridge gaps in expertise and perspectives, ensuring that the final digital solution meets all team goals, prioritising user satisfaction, usability, and utility. Additionally, building empathy through shared UX activities helps develop a more empathetic understanding of the end user. 5. Measuring Success With User-Centric Metrics Implementing human-centred design means going beyond technical success metrics and focusing on including metrics that reflect user satisfaction and performance: a) regular feedback from users on their satisfaction with the tool and specific features; b) task success rates and time to complete them; c) feedback loops such as surveys or regular feedback sessions that help identify and resolve issues in real time, rather than waiting for problems to escalate. Conclusion Moving from “human-in-the-loop” to human-centred design is a mindset shift that emphasises designing for users rather than simply using users to validate machine output. In companies where decisions can have a high impact and significant risk, human-centred design helps better align tools with users' real-world needs and context. For data scientists, this mental shift paves the way for building digital tools that are not only functionally efficient but also truly empower users and enable them to engage a broader audience. When data tools are built with a human-centric mindset, they become tools that help users make better, faster, and more confident decisions, ultimately improving both the user experience and operational outcomes. Data science and user experience design have recently become critical to the success of companies' sophisticated internal digital tools for analyzing, processing information, and predicting what-if scenarios. Used by internal teams and operations staff, these tools aim to optimize processes, predict outcomes, and inform decision-making in increasingly complex systems. For those working in data science, a common approach is to view humans as “human-in-the-loop,” meaning users provide inputs to a system or validate outputs, but their role remains secondary to technology and algorithms. It is worth considering that the complexity and nuanced requirements of modeling tools have begun to require a shift to a more human-centered approach to make the tools accessible to a wider audience. This article examines human-centered design and the need for such a focus for those working on data science projects and suggests strategies to facilitate this transition. Determining the Difference To address human-related issues (emotional state, learning, outcome), the concept of “human-in-the-loop” (HITL) was introduced, which involves incorporating human knowledge into the modeling process. The HITL approach is often a common way of thinking among data science teams, where humans are seen as secondary players who help the machine function better or validate the results generated by algorithms. This interdisciplinary approach crosses computer science, cognitive science, and psychology and consists of three components: data preprocessing, data modeling, and process modification to improve performance. However, it is important to consider that the results and performance of ML models can be unpredictable, so it remains unclear which aspect of human-machine interaction yields the best learning results. People are thus viewed as add-ons to the system—sources of labeled data, corrections, or ideas that help fine-tune the models—without taking into account their broader needs, motivations, or limitations. Human involvement depends on the type of solution being implemented: strategic, tactical, or operational. In operational and tactical systems, the human role is usually minimal. In contrast, in strategic solutions, the human role is critical because these decisions require context, explainability, and domain expertise. They often rely on human interpretation to validate assumptions, assess risks, and align outcomes with broader goals. Human-centered design, on the other hand, places the user at the center of the design and development process by aligning technology with human needs, behaviors, motivation, and goals. Understanding the "why" behind user behaviors is a key component of human-centered design, which integrates user context and requirements thoroughly into all phases of product development. For data scientists, embracing human-centered design is more than just incorporating human input into systems; it also entails actively reshaping systems to meet the actual users’ needs and developing tools that empower rather than burden them. Why Human-Centered Design Matters In complex, high-stakes environments, inconsistent forecasts, confusing workflows, or overly complex interfaces can lead to poor decision-making, frustration, and operational inefficiencies. That’s why a human-centered approach is invaluable in such settings, enabling users to receive answers tailored to their needs. Often, users view data science solutions as “black boxes,” making it difficult to trust them or understand the results they produce. Developing tools and solutions that explain the relationships between inputs and outputs and the behavior of models can improve trust and usability, empowering users to confidently make informed decisions. This need for explainability and transparency is directly related to the need for human-centered design and more understandable solutions. When tools are designed with end-users in mind, they are more likely to be adopted and used effectively by users. When data scientists understand the nuances of user tasks, they can create models and interfaces that match users’ real workflows and contexts, rather than idealized or abstract ones. This increases trust in the tool’s output and minimizes the learning curve for users. Moreover, the accuracy and applicability of machine learning models can be highly dependent on the user context. These subtleties are better captured by a human-centered approach, which ultimately produces more accurate and meaningful predictions for users. By providing users with the tools they need to make informed decisions, human-centered design empowers users. Instead of relying on the user to validate a machine’s predictions, tools can allow users to make decisions alongside the machine, fostering collaboration rather than a directive relationship. Practical Steps to Move Toward Human-Centered Design While the benefits of human-centered design may be clear and obvious, shifting from a “human in the loop” requires both structural changes and cultural and team support. Here are some actionable steps teams can take to embrace human-centered design. 1. User Research First Researching users and their context is fundamental to human-centered design. Thus, interviews, ethnographic studies, observations of users’ workflows, and identification of pain points, motivations, and goals help to better understand and study users. Such research produces qualitative insights that can be complemented by quantitative data, providing a deeper understanding of users' context. Additionally, creating user journey maps that detail the workflows and touchpoints users have with various tools and other organizational functions allows for a more accurate assessment of what users are experiencing. This visual tool can help highlight opportunities where these tools can add more value or streamline user processes. 2. Prototype With a Purpose To make efficient use of a team's time by experimenting with different models, assumptions, and data visualizations while focusing not only on achieving technical goals but also on creating features that are clear, relevant, and useful to the user, rapid prototyping can help. It’s a fundamental practice of human-centered design that brings ideas to life and allows users to test the prototype early, for example in Figma, without spending time on development. This ensures that user feedback is obtained before a huge amount of time is spent on product development and helps to quickly get to the point where the user's problem is solved by matching the real needs and defining what tasks the user needs to perform to achieve it. 3. Design for Trust and Transparency A critical component of human-centered design is building trust and transparency so that when users receive forecasts or recommendations, they understand the validity of those results and can make business decisions and act on them with confidence. Trust in data tools can be achieved by helping users understand why the system suggests a certain outcome or path and by visualizing results that are understandable to the user and consistent with their mental model. 4. Development of a Collaborative and Interdisciplinary Team Culture Creating successful human-centered data tools requires collaboration across disciplines. Having a shared team vision based on the human-centered design will help bridge gaps in expertise and perspectives, ensuring that the final digital solution meets all team goals, prioritising user satisfaction, usability, and utility. Additionally, building empathy through shared UX activities helps develop a more empathetic understanding of the end user. 5. Measuring Success With User-Centric Metrics Implementing human-centred design means going beyond technical success metrics and focusing on including metrics that reflect user satisfaction and performance: a) regular feedback from users on their satisfaction with the tool and specific features; b) task success rates and time to complete them; c) feedback loops such as surveys or regular feedback sessions that help identify and resolve issues in real time, rather than waiting for problems to escalate. Conclusion Moving from “human-in-the-loop” to human-centred design is a mindset shift that emphasises designing for users rather than simply using users to validate machine output. In companies where decisions can have a high impact and significant risk, human-centred design helps better align tools with users' real-world needs and context. For data scientists, this mental shift paves the way for building digital tools that are not only functionally efficient but also truly empower users and enable them to engage a broader audience. When data tools are built with a human-centric mindset, they become tools that help users make better, faster, and more confident decisions, ultimately improving both the user experience and operational outcomes.

Walkthroughs, tutorials, guides, and tips. This story will teach you how to do something new or how to do something better.

The writer was physically present in relevant location(s) to this story. The location is also a prevalent aspect of this story be it news or otherwise.

Key Design Principles for Startups: Things You Must Prioritize

Design and Data Science: From a Human-in-the-Loop Approach to Human-Centered Design

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

A Human-centred Framework for Designing Complex AI Solutions for Strategic Decision-making

9 Free Data Science Courses & Guides For Beginners

8 Skills Required To Become A Data Scientist

A Breakdown of Salaries of Data Scientists

A Step-by-Step Guide to Failing a Data Science Project

Automatic Feature Selection in Python: An Essential Guide

A Human-centred Framework for Designing Complex AI Solutions for Strategic Decision-making

9 Free Data Science Courses & Guides For Beginners

8 Skills Required To Become A Data Scientist

A Breakdown of Salaries of Data Scientists

A Step-by-Step Guide to Failing a Data Science Project

Automatic Feature Selection in Python: An Essential Guide

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps