ピープルアナリティクスの教科書 組織・人事データの実践的活用法

Review

"People Analytics" aims to contribute to the future of workers and management by visualizing and analyzing human resources data. This word has rapidly gained recognition in Japan since 2016. As a result, as of 2019, the percentage of companies planning to introduce or have already introduced it is as high as 51%. However, the reality is that many companies have not yet taken advantage of it, or have stopped at the testing stage.

This book is literally a "textbook" that the human resources managers of such Japanese companies should first pick up. The main targets are people analytics practitioners and managers, as well as the data analysts who support them.

Of course, when introducing such a new concept in-house, there are obstacles such as resistance forces such as old-fashioned groups and excessive "expectations and illusions" and the resulting "disappointment and disillusionment". The tips for overcoming such hurdles are also carefully written in this book. I think that will evoke the sympathy of many readers, not just human resources, and will support the introduction of people analytics.

In the field of human resources, where it is difficult to obtain quantitative data, management based on human experience and intuition was the mainstream. People analytics is not a replacement for it, but rather a strong support for human management. This book will be a powerful weapon for updating talent management.

The main points of this book

Point 1

People analytics is to improve the accuracy of decision making in human resources management, improve the efficiency of operations, and improve the value provided to employees.

Point 2

By combining the experience and intuition cultivated in human resources with data analysis and utilization, it will be possible to shift to data-driven human resources management.

Point 3

Personnel activities have issues such as comprehensive measures, a tendency to be a May rain ceremony, and difficulty in continuous efforts as an organization. However, the use of people analytics makes it possible to prioritize activities based on quantitative data.

What is People Analytics?

Improving decision-making accuracy

People analytics means "a method that utilizes various data related to human resources management to improve the accuracy of human resources management decision-making, work efficiency, and increase the value provided to employees."

One of the most obvious examples is the automatic decision of recruitment. This is a statistical prediction of whether or not a person (high performer) will demonstrate high performance after joining the company, based on the personality and ability characteristics data collected at the time of selection. It is called "a recruitment decision system that utilizes AI".

Speaking of recruitment selection, in the past, it was common to make a cut-off with application documents and tests related to personality and ability characteristics, and then make a final pass / fail decision by interview. In these selections, the guidelines presented by the company to the interviewer provide some criteria. However, in the end, there are many parts that rely on the interviewer's own "experience and intuition." People Analytics aims to improve the accuracy of these hiring decisions by providing data analysis suggestions.

[Must read point!] Tips for introducing people analytics

Recommendation of utilizing existing data

When actually promoting people analytics, there are many voices from within the Human Resources Department saying, "Human resources data is not well prepared and it is too early to analyze." It seems that many companies have not taken the first step due to this idea of ​​"priority on data maintenance".

Therefore, this book recommends a method of using existing data and leading the analysis, even if it is insufficient. By creating a quick win in this way, people around us will feel that the analysis of human resources data seems to be meaningful.

Of course, the analysis results may be unsatisfactory. However, once you analyze it, you can understand what kind of analysis your company needs in the first place, and you can narrow down the data you really want. Then you can start collecting new data.

Experience and intuition VS data analysis

It's easy to think that the purpose of people analytics is to shift from traditional “experience and intuition” -based talent management to evidence-based management. For this reason, experience and intuition are conflicting concepts with human resources data analysis and utilization, and many people are concerned that what has been cultivated in the past within the Human Resources Department will be denied.

However, in reality, experience and intuition are not totally denied. Rather, it will be possible to shift to evidence-based human resources management only when the two are integrated.

Hypothesis planning / interpretation / policy points

The key to data analysis is how to make a good hypothesis. "Hypothesis planning" can only rely on human insight based on experience and intuition.

Experience-based knowledge of human resources is also crucial in the process of how to "interpret" the results of data analysis and what "measures" to plan accordingly. In particular, when the result is unexpected and unexpected, new data may be buried if the reason is not explained properly. It is people, not data, who can explain it.

Furthermore, even if high performer data is obtained as an example, it is based on past data. Tuning measures for the future, such as future changes in business and the image of people needed, depends on the will of people and the ability to gain insight into the future.

Magic wand

In the world of marketing, when you analyze POS data, you may get a "surprise" that you are actually buying unexpected products together. Perhaps because of that, there are also over-expectations that there may be surprising discoveries in data analysis in the area of ​​human resources.

However, when we actually analyze human resources data, it is often proved that we felt something thin. Therefore, there are some voices from the Human Resources Department saying, "I knew that before I analyzed the data."

However, the results of such analysis are of great significance. This is because there are many measures in human resources that could not be implemented due to lack of solid evidence, even though they were slightly understood. It is necessary for the parties concerned to understand such things in advance.

What you can see with people analytics

Here are some of the most common challenges in Japan's Human Resources department. It is difficult to determine the priority of personnel issues. Therefore, as a result of working on the whole flower, it often ends up halfway due to lack of resources. Alternatively, as a result of starting with “easy-to-start” efforts such as low cost and completion within personnel personnel, the expected results may not be obtained.

Why do the HR departments face these challenges? That's because data has been scarce to prioritize measures. People analytics can be said to provide it.

For example, data analysis reveals the following: What percentage of the high performers in the company are likely to quit this year? What are the top three factors influencing high performer turnover? What measures will reduce the chances of leaving a job? When this becomes clear, it becomes easier to decide which countermeasure to start from based on quantitative grounds, and it becomes easier to reach consensus.

Basic procedure

Formulation of purpose / plan

When it comes to analytics, analysis methods using statistics and machine learning using AI are highlighted. But that's only part of the whole. From here, we aim to grasp the whole picture by following the basic procedure of data analysis.

Step 1 is the formulation of the purpose / plan. First of all, it is most important to embody "what you want to achieve", "what kind of action to take for that purpose", and "how data analysis contributes to that action".

For example, what about the purpose of "increasing the cost-effectiveness of recruiting activities"? An action of "increasing the proportion of high performers" can be considered, and an action of "reducing the cost for forming a recruitment population" can be considered. Depending on that, the data to be analyzed will be different.

Identifying and collecting necessary data

Step 2 is to identify the necessary data. For example, in order to know the characteristics of human resources who are active after joining the company, there are data showing the characteristics of human resources and data showing the active status after joining the company. The former is data such as "appropriate test" and "entry sheet" at the time of hiring. For the latter, data such as "performance evaluation," "personnel evaluation," and "multi-faceted evaluation" after joining the company are candidates.

Next, in step 3, data is collected. In many cases, necessary data is stored separately in each department, such as the recruitment section for data at the time of hiring and the personnel planning section for data after joining the company. Therefore, it is necessary to collect and unify the data with the permission of data utilization.

Process data

Step 4 is data preprocessing. The data collected is often not ready for immediate analysis. Full-width and half-width characters may be mixed in individual attribute information, and evaluation symbols with different definitions may be used for each year due to changes in the evaluation system. It is necessary to prepare data in such a non-uniform format. Such correction / processing of data is called "cleaning" or "cleansing".

Preprocessing data is a steady and time-consuming task. However, since it is an indispensable procedure for the effective and correct use of data, it is necessary to pay close attention to it.

Data visualization and analysis

Step 5 is data visualization. Understand the basic characteristics of the data using screening methods such as "frequency distribution" and "scatter plot".

In step 6, the analysis is finally carried out. After going through the process up to this point, we can finally perform "analysis" such as statistical analysis and machine learning. Due to the variety of analytical methods, it is not practical for a person with a human background to understand all of them in detail. Therefore, it is necessary to collaborate with experts in statistical analysis and machine learning, and to acquire the minimum knowledge that is the basis of dialogue.

Interpretation and implementation of measures

Step 7 is a decision. After the analysis is finished, be sure to check the prediction accuracy. If the accuracy is low, it is necessary to decide the subsequent policy such as whether to use the prediction model, redesign the model, or give up the prediction. In addition, if a result contrary to expectations is obtained, it is necessary to carefully consider the result while confirming the accuracy and mobilizing all of one's own knowledge and experience.

The final step 8 is the development of measures. If the direction of future measures can be derived from the analysis results, the measures will be actually developed. If it is a high performer's forecast, use the result to prioritize recruitment or review recruitment requirements. It can be said that the analysis is completed only after taking such actions.

Learn from case studies

Interviewer's connoisseur

Let's introduce the case of Uber from the case of people analytics in this book.

When hiring a large number of employees, the company focused on the number of hires as well as indicators such as "whether the hiring people performed well after joining the company" and "whether the hiring people became established". With these, they tried to measure the quality of recruitment. Therefore, we analyze the connoisseurship of hiring interviewers by linking the candidate database at the time of hiring with the employee database after joining the company. As a result, it was found that 90% of the employees hired by interviewers with the top 10% of connoisseurs became top performers after joining the company. On the other hand, 90% of the employees hired by the bottom 10% of interviewers were found to have left their jobs within a year of joining the company.

So Uber decided to remove the bottom 10% of interviewers from the interview process and, conversely, get the top 10% of interviewers involved in the recruitment process more extensively. By practicing people analytics in this way, the accuracy of the company's recruitment has improved dramatically.

Recommendation of reading

In this book, specific cases of nine companies are introduced as "Case Studies". If you are new to people analytics, it may be easier to understand if you start reading here. In addition, at the end of the book, materials related to "Principles for Utilizing Personnel Data" are summarized as "Materials". These will be valuable guidelines in practice.

Also, if you are a member of the HR department and management, please read Chapter 6, which describes the organizational response to operations, regardless of your duties. It can be very useful for laying the foundation for the correct and effective introduction of people analytics.

おすすめの記事