where you are data driven it has become a badge of loyalty. Organizations proudly talk about the dashboards, AI techniques, predictive models, and automation they've invested in and benefited from. As the internet can tell you, almost every Fortune 1000 company is increasing its investment in data and AI to stay agile and competitive. And yet, without unprecedented access to the quality and quantity of data, many analytics and AI programs never reach production or are unable to make a lasting impact.
Data models are built, information is shared, decks are applauded and then quietly forgotten only to be (what I like to call) junk boards.
In this day and age of machines taking over our decision-making power, the problem isn't a lack of data, talent, or tools – it's of a person that we are starting to forget to talk to him.
This is where it is Human-Centered Data Analytics It is not only appropriate, but important.
What is a People-Centered Approach?
Data is nothing but digital traces of human interaction. A human-centered approach can improve the data choices that scientists make every day, by making the process more transparent, questionable, and consider the social context of the data.
A person-centered approach asks a simple question:
Who is this for and how exactly will it be used?
Now think about it this way–in asking “What can we predict from this data?”A person-centered approach makes us want to ask “What should we help people understand or decide with this data?”
Human-Centered Data Analytics is the concept of understanding how people interact and make sense of social situations, allowing people to explore and discover data, and design data models with the end user (not just the business) in mind.
At its core, human-centered Data Analytics means designing models and metrics with the end user in mind, not just a business KPI. It asks us to improve the day-to-day decisions that data professionals make: how we pose problems, what features we create, what metrics we prepare, and how we communicate solutions to those problems.
Why Human-Centered Data Analytics is the Future
As the world becomes more technological and business driven, we as a society are shrinking social relations and behavior. Organizations, no matter what type of business they are, bring people down to make profits and opportunities. We forget that every dataset comes from someone who decides to buy, click, submit, vote, or opt-out and end up treating this behavior as a signal instead of a story.
Ignoring that human context can lead to making the wrong result entirely. A person-centered approach introduces a new dimension and forces us to ask:
- Who benefits from this model?
- Who might be harmed?
- What assumptions are made about the data?
How to Practice Human-Centered Data Analytics in Your Work
My tendency toward a person-centered approach is not a new passion.
Early in my career, I was deeply interested in Human–Computer Interaction (HCI)—the field that studies how people design, use, and interact with technology. Through working with HCI, without much awareness, I developed an attitude of prioritizing human understanding, behavior, and social context when solving a problem.
So even though I'm in the field of data and AI now, a human-centric mindset has become second nature to me. Over the years of working as a senior analytics consultant, incorporating a Human-Centered approach has only required simple, intentional shifts in the way I work and this is how I implement Human-Centered Data Analytics in my work.
1. Start with People, Not Metrics
In the early years of my career, my mind was set on designing beautiful dashboards because that was the tangible result that made me visible. However, over time, as I grew as a data professional, I realized that dashboards do not create value by themselves. Decisions are made.
You need to design your analytics around the decisions people can make from the analytics, not just from the dashboards. Before defining any measures or KPIs for your analysis or dashboard, you should ask:
- Who could use and act on these ideas?
- What decision are they trying to make?
- What challenges do they face?
Asking these questions to people who have been affected before often explains the next steps for me, removes the guesswork and ensures that the metrics I'm sharing help the problem, instead of hoping that the metrics I have are true for the problem I'm solving.
2. Investigate the Root of the Problem
Every problem has a history.
Human-Centered Data Analytics asks us to think about the questions relevant to the problem and pause before collecting, crunching, and manipulating the necessary data. You should document assumptions and known biases, not just as footnotes, but as part of the analysis. Ask questions like:
- Where did the problem come from? Under what circumstances?
- What behaviors are missing or underrepresented?
- What data can answer this problem in the context asked?
This creates transparency and sets realistic expectations for how information should be interpreted.
3. Design for Understanding, Not Just Accuracy
A data model with 94% accuracy that no one understands is rarely going to make an impact.
But, if you pair the output from that data model with a short narrative that explains why the result exists, not just what it is, see for yourself how that makes an impact. Human-centered analytics pushes you to translate technical language into simple human understanding.
Once your data model is ready, ask:
- Can a non-technical participant explain your information after hearing it once?
- Can you replace feature importance charts with decision-oriented visualizations (eg, “If X goes up, here's what changes”)?
- Can you trade the gain of less precision for clarity?
A human-centered approach allows you to design models with improved detection and accuracy.
4. Account for What the Data Does Not See
I can't stress enough how much this has allowed me to grow in my career! Being able to see the short coming of the dataset, anticipating questions about those gaps and preparing to answer that gap has been the main driver of my promotion up the ladder.
But hey, no points for guessing where that comes from – a human-centered approach to working with data!
A person-centered approach allows you to clearly acknowledge blind spots. As you become familiar with the dataset, start documenting known data gaps, behavioral patterns of the dataset, and announce assumptions during the presentation instead of leaving them vague. You may ask:
- What does this data not show?
- What group or behavior is underrepresented?
- Can the judgment made by decision makers on these data points stand if the gaps are significant.
4. Design for Behavioral Outcome, Not Just Performance
Working with sensitive data makes ethics unavoidable. But because of the human-centered approach, it allows us to treat ethics as a design constraint, not a compliance check box. Ask ethical questions early and plan for them, not as an afterthought, such as:
- What happens if this data model is not appropriate?
- Who will bear the cost of errors?
- How will the answer be put together?
By planning for these situations in advance, I can create solutions that are not only successful, but responsible and sustainable.
5. Build Feedback Loops in the Program
As part of the workforce, we all know the importance of feedback and integrating that into our work and not just from a data perspective, but a holistic, human-centered approach pushes me to treat solutions as evolving programs rather than one-time deliveries.
From a human-centered approach, your design for adding feedback loops to your systems is a 3-step process:
- Define success metrics beyond the launch (such as acquisition, rollout, and stakeholder confidence)
- Schedule recurring check-ins with users and stakeholders to understand how information is being used or ignored
- Include quality feedback in future iterations, not just quantitative performance metrics.
The results from step 2 above on how information is used or ignored may not always be what you wanted. I hear a lot of “oh we don't use that tool anymore” from tools I've built in the past. So to avoid that, keeping a human-centered approach in mind, ask questions before and after the creation of the tools-
- How will this analysis be evaluated and used once implemented?
- Should this be a one-time delivery or a robust tool?
- How many users have stopped using the tool only after a few uses? What has changed?
Closing Thoughts
Data Is Powerful Because People Are.
The future of math isn't about more data, bigger models, or faster pipelines—it's about intelligence!
Human-Centered Data Analytics reminds us that data is powerful not because it has meaning, but because it reflects human life in all its complexity. When we design math with empathy, context, and responsibility, we not only build better models but better systems!
And that is more important than ever.
Tmy goal in this blog post. Thanks for reading! I hope you found it an interesting read and have a great time this new year telling stories with data!
Rashi is a data wiz from Chicago who loves analyzing data and creating data stories to communicate insights. He is a full-time healthcare analytics consultant and likes to blog about data on the weekends over a cup of coffee.
