Summary of Articles on People Analytics
I have been managing a magazine on note called "People Analytics Toolbox". It is a compilation of my experience in HR data analysis, which originally started as a way to answer questions from members of our internal HR department and analysis team.
In this article, I would like to organize the contents of the magazine and summarize them according to analysis scenarios. I will also post updates here.
Updated 2026/4/10
Added "How to introduce people analytics?" and "Further information sources".
Revised some articles and added/changed introductory articles.
Changed the eye-catching image.
What is people analytics in the first place?
This is the very first article I posted in the magazine. At the time, we called it HR analytics within the company, so it has this title.
This article answers the question of what HR data analysis is in the first place. Here, I defined it as follows.
The primary initiative of HR analytics (people analytics) is to support quantitative decision-making based on facts for HR measures.
This definition is based on my own experience and is still the axis I use when explaining it to clients. On the other hand, I feel that the scope of people analytics utilization is expanding. Specifically, I believe it has the following benefits.
Enable decision-making based on facts rather than intuition and experience.
Incorporate hypothesis-driven work processes into HR to realize strategic HR.
Realize operational efficiency within the HR department.
Enhance communication between employees and the HR department.
Lead to organizational development within the HR department.
The first point above was mentioned in the previous article, but I have written about the others in separate articles, so let me introduce them in order.
People analytics to realize hypothesis-driven HR
This article describes the relationship between people analytics and OODA, and is based on insights gained during advisory work for HR departments.
At the time, there were several people analytics projects running within the HR department, but we faced challenges in launching these analysis projects.
Data analysis does not work without a hypothesis to verify, but the very act of formulating a hypothesis was considered difficult in the first place. This is a challenge that I still find resonates with clients today.
Therefore, I developed and rolled out a workshop specifically for setting themes in people analytics. During this, one HR executive said the following:
People analytics brings hypothesis testing into HR. However, when you think about it, isn't this a basic practice that HR should have mastered before even talking about data analysis?
Hearing these words, I realized that was exactly right. Thinking about the gap between the goal and the current situation, formulating a hypothesis to solve the problem, and executing it—these are necessary in any line of work. And I realized that HR, where measuring the effectiveness of measures is difficult, is full of potential for data analysis.
I summarized the insights gained here and framed the hypothesis testing process as the DDDI cycle. Please take a look at this as well.
For how to think about hypotheses and analysis themes, please see the next article.
A Framework for Thinking About Analysis Themes (TIHAM)
When it comes to HR data analysis, people tend to focus on the methodology of collecting data and looking at it with Tableau or R. However, what is important in data analysis is the purpose: why are we analyzing, what do we want to know through the analysis, and what actions do we want to lead to?
Methodologies themselves, such as data analysis, machine learning, and AI, are fascinating and I love them too, but analysis without a purpose will not lead anywhere.
Some might think, 'Since we are putting effort into the analysis, there is no way we don't know the purpose.' However, the more advanced the application field, the more common it is to confuse means and ends.
To minimize this problem, I developed a framework for organizing themes before starting people analytics work. I'm not sure if it's easy to remember, but I call it TIHAM.
This framework is a simple one that organizes 1) Purpose, 2) Issues/Ideas, 3) Hypotheses, 4) Approaches, and 5) Measures, but I believe it is quite powerful. In particular, it is important that you can think about the measures to be taken after hypothesis testing before you even start the analysis.
Now, as a tip for considering such themes, there is a technique of 'imagining what happens after the analysis before starting the analysis.'
If you cannot imagine the action to be taken after obtaining the results of the data analysis, then that analysis might not be necessary.
Perspectives on People Analytics
Once the purpose and problem definition of the data analysis are decided, you will need to decide on an analysis approach. At this time, you may be unsure about what perspective to use when looking at the data. In the article above, I have roughly organized those perspectives and summarized them in a mind map.
For example, if you are analyzing the experience of high performers, you might think about combining performance data with activity data and training history. At this time, it is very important to understand what kind of data can be utilized in the first place.
Some people often say that 'it is difficult to analyze with only HR data,' but if you include data from related systems, the information within an organization is quite extensive.
Please use the mind map as a reference. You will surely find a perspective that you think is the one.
The Rather Troublesome Preprocessing of HR Data Analysis
The biggest hurdle in practicing people analytics is data preprocessing. This is because data scattered across various systems must be aggregated, integrated, and organized into a format suitable for analysis. The article above discusses points to be aware of regarding preprocessing. It may also provide hints for designing a data integration infrastructure.
While it would be ideal to have a dedicated data warehouse or data mart for HR data, this takes time and money, so you will likely start analyzing while the infrastructure is still being built. I was in that position once myself.
When performing data analysis in a completely untouched field, you generally struggle with data preprocessing, but the HR field has unique challenges hidden within it.
In a word: history hell and code hell.
I worked on developing HR systems during my time as a software engineer, so I had a knack for it, which helped, but those who haven't will likely be bewildered. I think you can get an idea by looking at the article above.
How do you go about the analysis?
Once data preprocessing is finished, you finally move on to analysis. What is required of an analyst is to apply and evaluate analytical methods according to the problem setting, gain insights, and report them.
The following article introduces the flow of analysis using engagement analysis as an example. I have also made toy data (demo data) available, so please take a look.
Also, for those who want to know about realistic analysis cases, please see the following article.
On the other hand, I am surprised that we are now in an era where we can perform analysis based on instructions using generative AI, but in order to properly provide those instructions and evaluate the results, a person must be well-versed in analysis. Therefore, those who practice people analytics need to learn analytical approaches through various experiences and self-study.
I once became a data scientist without any experience, and I ended up in a difficult situation. I struggled through it as a late bloomer in my 30s, but the world is much easier to learn in now than it was back then.
People Analytics for Improving HR Department Work Efficiency
Up to this point, I have been thinking about HR data analysis as decision-making support, but it can also be used to improve operational efficiency. An easy-to-understand example is the task of aggregating overtime work status by department and passing it on to those departments. This point is discussed in "Scene 2" of the article above.
Such tasks are routine for HR and general affairs, but they take more work than you might think.
Perhaps it's something like pulling data from HR or attendance systems in some way, grinding away at it using Excel, storing the files on a shared drive or in the cloud, and notifying people via email...
To make this more efficient, it is necessary to automate all or part of the five processes: (1) data aggregation and processing, (2) data tabulation, (3) data visualization, (4) sorting for departments, and (5) notification to departments.
It might be difficult to automate everything all at once, but for example, if (1) and (2) are the most time-consuming parts, a quick solution is to make those parts a batch process. Alternatively, processing (2), (3), and (4) with a BI tool like Tableau on your local machine would also be effective. To realize (5), it becomes necessary to organize data governance.
Such discussions are a methodology that is effective not only for attendance but also for feedback on internal surveys.
People analytics leads to organizational development within HR
The work of the HR department is diverse, covering recruitment, placement, evaluation, talent development, payroll and attendance, and welfare, and it can sometimes be siloed. What I have felt through practicing people analytics is that it is necessary to jump over these barriers to conduct effective analysis.
For example, suppose you are discussing issues related to improving performance for a specific job type within the company. In this case, it is not uncommon for discussions to extend beyond talent development to mid-career recruitment and placement as human measures for performance improvement. And it is often the case that opinions from each team intersect and do not come together.
In such situations, what brings focus to the members is 'facts.' When there is a discussion about whether this or that is the case, you look at those assumptions using data. And I have seen many times how this becomes a trigger for discussions to move forward positively.
The story so far should not necessarily be limited to data, but should also utilize employee interviews and the like. However, by having the perspective of people analytics, a viewpoint for ensuring objectivity regarding interviews is born.
In this way, I believe that the activities of people analytics themselves will have some kind of positive impact on HR, and eventually lead to organizational development in HR.
How to introduce people analytics?
So, if you are actually going to start people analytics, where should you start?
What I recommend to clients is to 'catch the little questions that arise in daily meetings and look at the data.'
While I would like you to have basic knowledge of statistics, what I recommend to clients is a simpler story: 'save your daily questions.' 'Isn't engagement among people in their 30s dropping?' 'I feel like the turnover rate in that department is high.' I think things like that are discussed on a daily basis. Why not start by verifying them one by one using data?
People analytics is just one of the means for considering measures and making decisions. I believe that its source actually already exists in daily work.
If people analytics helps with business decision-making, I believe it is effective to internalize it. However, it also takes time to introduce from both technical and cultural aspects. It does not mean that it is okay just to have analysis skills.
I support the introduction of HR data analysis teams while providing side-by-side support for the launch and execution of real projects. Please feel free to consult with me.
Further sources
I have compiled a new list of recommended books on people analytics!
I introduce a wide range of books, from data analysis technology to HR work, so please take a look.
I have launched a site that summarizes practical knowledge of people analytics. It aggregates technical blogs and newsletters from Kuni-Lab.
Come to think of it, this is an article that looks back on why I am doing people analytics. Please use it as a reference.
Related article: