People analytics is the analysis of data about the people in an organization to solve business problems: why talent leaves, which profiles perform best, or where successors are needed. In a medium-sized company, it doesn't require a large data team, but rather a clear question, organized data, and privacy rules from day one.
Table of Contents
What is people analytics and how does it differ from HR reports?
The CIPD, The British Institute of HR Professionals defines it as analyzing data about people to solve business problems, and notes that it is also called HR analytics, talent analytics, or workforce analytics. If you are already familiar with the term from our article on HR analytics, Here we focus on how to get it up and running.
The difference with a report lies in the question. A report states how many people left last quarter. People analytics seeks to understand why they left, whether there is a pattern, and what the best course of action should be. John Boudreau and Wayne Cascio point out in an article from 2017 that most organizations still have difficulty moving from operational reporting to analysis.
That doesn't mean indicators are superfluous. They're the foundation. If you haven't defined them yet, start with the HR KPIs with formula and responsible.
What levels of analysis exist?
The CIPD distinguishes three levels. These help you understand where you stand and what it's reasonable to ask of your team.
| Level | Question that answers | Example in HR. | What do you need |
|---|---|---|---|
| Descriptive | What happened? | Rotation by area and seniority in the last 12 months | Clean data and a dashboard |
| predictive | What will happen? | Next year's staffing needs based on historical data and business plans | Sufficient historical data and statistical analysis |
| Prescriptive | What should we do? | Recommend courses to each employee based on their skills gap | Validated models and rules of use |
According to the CIPD, most organizations can perform descriptive analysis, but few reach the prescriptive level, and using advanced analytics doesn't always yield more valuable information. For a mid-sized company, a well-interpreted descriptive analysis is often more effective than a predictive model lacking supporting data.
What data does people analytics use?
The data comes from HR systems, other company systems, and external sources, such as salary surveys. This table summarizes the common sources and their uses:
| Font | Examples of data | What is it for? | Careful |
|---|---|---|---|
| Payroll and HR system. | New hires, terminations, position, seniority, salary | Turnover, personnel costs, pay equity | Financial and asset data |
| Recruitment | Vacancies, stages, sources, times | Hiring time, conversion by stage | Keep only what is necessary from someone who was not hired |
| Assessment | Results of skills and competency tests | Compare candidates and measure gaps | Use them for the informed purpose |
| Performance | Assessments, objectives, 360 feedback | Relating performance to selection and development | Evaluator biases |
| Climate and surveys | Commitment, intention to stay | Anticipate turnover and prioritize actions | Anonymity and minimum group size |
| Training | Courses, hours, certifications | Measuring the effect of training | Do not confuse attendance with learning |
| External | Salary surveys, labor market | Compare salaries and plan workforce | Job comparability |
You don't need to have everything. The practical thing to do is to cross-reference two or three sources that answer a specific question.
What is the purpose of people analytics in a medium-sized company?
These are the use cases with the highest return on investment when starting out:
Rotation and retention
The most common scenario, and the one the CIPD uses as an example, involves identifying groups with high turnover and adapting incentives to reduce it. This involves cross-referencing voluntary resignations with department, manager, seniority, salary, and climate survey results. Often, the pattern is found within a specific team or at a particular time, such as the first six months.
Recruitment quality
If you use tests in your selection process, people analytics allows you to verify with your own data whether people who scored highly in a competition perform better a year later. This helps you decide which tests to keep and how much weight to give them. For this comparison to be meaningful, the tests must provide... comparable results between candidates and positions.
Talent and succession map
Cross-referencing performance, potential, and skills reveals who is ready for a key position and where there is a risk of being left without a replacement. This is the logic behind it. Talent Mapping and tools like the 9-box matrix. At Hirint, assessment results feed that map with skills data, and the platform is ISO 27001 certified and operates under the GDPR.
Performance and development
Linking performance evaluations to training, mobility, and seniority helps to see which development actions translate into improvements and which do not.
Pay equity
Comparing salaries by position, level, and performance allows for the detection of gaps that cannot be explained by the work itself. It is a sensitive analysis: it should be conducted in conjunction with the legal department and with clear access rules.
What does Mexican law say about privacy?
People analytics works with the personal data of employees and candidates. In Mexico, the framework is the Federal Law on the Protection of Personal Data Held by Private Parties, published as a new law in March 2025. Some points that directly affect an analytics project:
- Consent and privacy notice. All processing is subject to the data subject's consent, except as otherwise provided by law (Article 7). Financial or asset data require explicit consent.
- Sensitive data. Health, ethnic origin, religious, philosophical and moral beliefs or political opinions, among others, require express written consent (Article 8).
- Purpose. If you want to use the data for a purpose other than that stated in the privacy notice, you need new consent (Article 11). Using selection results for a rotation model could be considered a new purpose.
- Automated decisions. The data subject may object to automated processing that assesses, without human intervention, aspects such as their professional performance and significantly affects their interests (Article 26).
In addition, the ICPD recommends the following best practices: be transparent about what is analyzed and why, consult with employees, ensure that the analysis does not discriminate against underrepresented groups, and appoint a data protection officer. This is not legal advice: review each project with your legal department.
How to get started with people analytics step by step?
- Choose a business problem. The CIPD insists that solving a critical business problem is what creates the most value. For example: "We lose too many salespeople before they reach their first year.".
- Conduct a data audit. Identify what data you have, where it is located, and its quality. A poorly captured "reason for leaving" field invalidates the analysis.
- Organize the project with a frame. Boudreau and Cascio use the LAMP model: logic (what relationship do you expect to find), analysis, measures, and process (how will the finding reach the decision-maker).
- Start with the descriptive. A segment-based rotation board, reviewed monthly, already allows for action.
- Define access and privacy rules. Who sees what, with what level of detail, and for how long the data is kept.
- Present findings with context. Unexplained data generates more questions than solutions. Accompany each finding with a proposed action.
- Measure the effect and repeat. If you changed something because of a discovery, check months later if the indicator has moved.
For the first project, you don't need to buy a platform. A well-structured spreadsheet or a business intelligence tool is enough to validate the question before investing.
Common mistakes when implementing people analytics
- Start with the tool, not the question. Without a question, the board fills up with data that nobody uses.
- Confusing correlation with causation. The fact that two variables move together does not prove that one causes the other.
- Making decisions about people based solely on one model. Data provides guidance; the decision about a person is made by another person, with context.
- Analyzing groups that are too small. In teams of few people, a result can identify someone even if their name does not appear.
- Forgetting the quality of the data. Poorly coded positions or inflated evaluations lead to erroneous conclusions.
- Not telling employees what is done with their data. The lack of transparency destroys the trust that the project needs.
Frequently Asked Questions
What is people analytics in a nutshell?
It's the use of data about an organization's people to solve business problems, such as turnover, hiring quality, or succession planning. It combines data from HR, other company systems, and external sources. Its value lies in turning that data into decisions, not in simply accumulating metrics.
What is the difference between people analytics and HR analytics?
In practice, they are used interchangeably, along with talent analytics and workforce analytics; this is how the CIPD defines them. Sometimes "people analytics" is used to emphasize that the analysis combines people data with business data. The important thing is not the name, but that the analysis answers a specific question and leads to a decision.
What are 5 important KPIs in human resources?
It depends on the strategy, but five common metrics are voluntary turnover, hiring time, cost per hire, absenteeism, and new hire performance. Each requires a fixed formula and a designated person responsible. People analytics starts with these indicators and seeks to explain their variations and what to do about them.
Do you need a data scientist to get started?
Not for the first project. A well-designed descriptive analysis can be done by someone in HR with good spreadsheet skills and support from finance or IT. Predictive models do require statistical knowledge, sufficient historical data, and validation. It's best to develop them once the descriptive analysis is already working and in use.
Can people analytics make decisions about employees?
He shouldn't make decisions alone. Data helps detect patterns and prioritize, but decisions about a person need context and human judgment. Furthermore, in Mexico, the individual can object to automated evaluations of their professional performance without human intervention when these evaluations significantly affect them, according to applicable regulations.
If you want to incorporate comparable skills data into your talent map, at Hirint we can review with you where to start.