HR Analytics

What HR analytics means

HR analytics (تحليلات الموارد البشرية), also called people analytics, is the use of employee data to answer management questions with numbers, instead of relying on impression or personal experience alone.

The word “alone” is part of the definition. Impression and personal experience are not excluded from a management decision; what HR analytics replaces is a decision that rests on them and on nothing else.

The four levels of HR analytics

HR analytics is not a single activity. It works at four levels, and each level asks a different question of the same data.

1. Descriptive. It asks what happened. The answers at this level are figures such as the number of employees, the turnover rate and the absenteeism rate.
2. Diagnostic. It asks why it happened. The answer is sought by breaking the figure down: in which department, in which period, and in which group of employees.
3. Predictive. It asks what is likely to happen. It requires a sufficient history of clean data, because an estimate of what comes next can only be built on what has already been recorded, and recorded correctly.
4. Prescriptive. It asks what should be done. It is the hardest of the four to apply, because it assumes that the three levels before it are correct. A recommendation built on a wrong description, a wrong diagnosis or a wrong estimate carries each of those errors into the action it recommends.

The descriptive level on its own does not answer a decision question. A report that shows what happened is useful, but it does not say what ought to change, and it does not say why the figure is what it is.

The condition that comes before any HR analytics

Analysis does not repair what the data has spoiled. HR analytics works on the data it is given, and if that data is flawed, the flaw passes into every result calculated from it. The following defects invalidate the results before the analysis has even begun.

• Definitions that are not unified. One department counts its trainees in its headcount and another department does not, so the turnover rate cannot be compared between the two. Each department has produced a figure, but the two figures are calculated on different groups of people, and setting them side by side compares two different things under one name.
• Data held in scattered places. Attendance is kept in one file, payroll in another, and departures in email. Before any question can be answered, the records have to be brought together, and each place where they are kept separately is a place where one record can disagree with another.
• Missing dates. They spoil any calculation that depends on time, such as average tenure. A period cannot be measured where its start or its end has not been recorded, so an employee whose joining date or leaving date is missing cannot be counted correctly in any measure of how long people stay.

The privacy limit on HR analytics

Employee data is personal data. The way it is handled is a question of employee data privacy, and it is subject to the provisions of the Personal Data Protection Law (نظام حماية البيانات الشخصية) in the Kingdom, issued by Royal Decree M/19 of 9/2/1443H and amended by Royal Decree M/148 of 5/9/1444H. Among the duties that law places on the controller, the party that determines the purpose and manner of processing, are the following three.

• Specifying the purpose of collection. Four articles of the Personal Data Protection Law set it out. Article 10 of the Personal Data Protection Law permits personal data to be processed only to achieve the purpose for which it was collected, subject to the seven exceptions that article lists. Article 11(1) of the Personal Data Protection Law requires the purpose of collecting personal data to relate directly to the purposes of the controller and not to conflict with any provision established in law. Under Article 12 of the Personal Data Protection Law, the controller must adopt a privacy policy, available before collection, that specifies the purpose of collection among its other elements. Article 13(2) of the Personal Data Protection Law requires that, where personal data is collected directly from the person it concerns, that person be told the purpose of collecting it.
• Limiting the data to the minimum that serves that purpose. It is set by Article 11(3) of the Personal Data Protection Law, which requires the content of personal data to be appropriate and confined to the minimum necessary to achieve the purpose of collecting it, while avoiding content that would lead to the person being identified specifically once that purpose has been achieved, and which leaves the controls needed for that to the Implementing Regulations of the Law.
• Protecting what is collected. It is set by Article 19 of the Personal Data Protection Law, which requires the controller to take the organisational, administrative and technical measures necessary to protect personal data, including when it is transferred, in accordance with the provisions and controls that the Implementing Regulations of the Law set out.

Alongside the text of the law, there is a practical rule within HR analytics itself: the analysis is carried out at the level of the group, not of the individual, and access to the data is granted, within defined limits, to those who need it. That rule is a practice in how HR analytics is conducted. It is not one of the statutory duties named above.

A methodological error in HR analytics

A correlation between two indicators does not mean that one of them is the cause of the other. This holds whether the correlation is seen in a table or estimated through regression analysis. When two figures move together, the movement shows that they are related in the data, but it does not show which of them, if either, produces the other.

Take a department that works longer hours and also has a higher turnover rate. The two indicators rise together, yet the higher turnover may come from the nature of the work in that department itself, not from the hours.

For that reason, the result of HR analytics is taken as a question to be verified, not as a decision ready to be applied. The result points to where to look. Whether the relationship it shows is a cause still has to be checked before anything is changed on the strength of it.

This is an explanation of the concept and of the statutory provisions cited, not legal advice.