Attrition Root Cause Analysis

What attrition root cause analysis is

Attrition root cause analysis is a method that traces employees’ departures back to the factor that actually produced them, rather than stopping at the reason an employee gave on leaving. Its output is not a description of the departures. It is one or more factors that the organisation has the power to change.

Attrition in this sense means people leaving by their own decision. The same word also names a way of reducing headcount by leaving vacated posts unfilled, which is a different concept, explained under attrition. Root cause analysis is concerned only with why people left.

Stated reason and root cause in attrition root cause analysis

The stated reason is what goes into the resignation letter or is said in the interview: “a better opportunity”, “personal circumstances”. It is true as a statement, but it cannot serve alone as the basis for a decision, because it describes the employee’s final decision and not what drove them to it.

The root cause is the factor whose change would have changed the decision. One test separates the two: ask whether the organisation has the power to change the factor. “A better opportunity” is outside its control. The gap between a job’s pay and the market for that job, the shift rota and the relationship with the direct manager are all within it.

How attrition root cause analysis is carried out: five steps

1. Define the case precisely. Which job, which department, what length of service and which period. An analysis run on all resignations at once ends in a general description.
2. Collect facts, not impressions. Dates of joining and promotion, changes of manager, pay adjustments, transfer requests and refused leave.
3. Build a timeline for each case, then look for where the cases meet. Cases that converge on a single event deserve to be followed up, even when their stated reasons differ.
4. Ask “why” until you reach a factor that can be changed, and stop at the first factor that an internal decision controls. Going further ends in general statements about the market.
5. Test the cause against cases that did not end in resignation. If the same factor is present among those who stayed, the cause is incomplete and needs a second factor to explain the difference.

Two errors that undermine attrition root cause analysis

The first is relying on a single case. One case reveals a possibility but does not establish a cause, and a decision based on it changes a whole system because of one event. How many cases are enough to establish a pattern is a separate question. In the sources we reviewed, we found no published rule on that number on which to base a threshold. What can be said is that one case is not enough.

The second is stopping at the first plausible cause. Plausibility is not evidence. The cause worth building on is the one that has survived an attempt to refute it with a case that contradicts it.

The same caution applies to ranking the causes of resignation by their strength. A ranking of that kind is a claim of measurement, and in the sources we reviewed we found no measurement on which to base one, so no order of causes is proposed.

How an overall rate hides concentration in attrition root cause analysis

The analysis can start from a single figure for the whole organisation, and that figure does not show where the problem lies. Take an organisation of 200 employees from which 24 left during a year:

• Overall leaving rate: 24 ÷ 200 = 12%.
• A department of 20 employees, 9 of whom left: 9 ÷ 20 = 45%.
• The rest of the organisation, 180 employees, 15 of whom left: 15 ÷ 180 = 8.3%.

The figure of 12% describes neither situation. It is an average between one that calls for attention and one that does not, and anyone who looks at it alone ends with a decision applied to 180 people because of what happened among 20.

The same arithmetic works in the opposite direction. A rate of 12% also appears if one or two people leave each department with no concentration anywhere. The two situations share the figure and differ in everything after it, because the first points to a local factor and the second to a factor that runs through the whole organisation. That is why defining the case precisely is the first step, not the last.

Concentration shows in time as well as in place. If 7 of the 9 had left within eight consecutive months, the question becomes what happened before those months, not what happens in the department in general. The third step asks for exactly this timeline, and it means nothing if the view stays at the level of the whole organisation.

One caution applies to the calculation itself. A rate calculated on a small number can move a great deal for reasons unrelated to what is being investigated. Two departures from a department of 6 people produce 33%, a figure that looks alarming and cannot bear a decision on its own.

The figures in this example were chosen for illustration and are not taken from any measurement. They show how an arithmetic mean hides concentration; they do not propose a rate to measure against. In the sources we reviewed, we found no published measurement on which to base a benchmark turnover rate for a sector or for a size of organisation.

How attrition root cause analysis differs from the tools beside it

Attrition root cause analysis is mentioned alongside other tools that share its subject but ask a different question. Whoever uses one of them in its place gets an output of another kind and then builds on it a decision it cannot bear.

• The exit interview. It produces one person’s account at the moment of leaving. That account is a useful input, but its source has an interest in ending the conversation smoothly, so such accounts are collected and not relied on alone. Root cause analysis is what examines ten of them together, looking for where they meet.
• The turnover rate. It is a measure that says how many, not why. It opens the analysis but does not replace it, and the calculation above shows that, looked at on one level only, it can close the question instead of opening it. Markov analysis takes the leaving rate as an input and does not explain it; explaining it is the work of root cause analysis.
• The employee opinion survey. It asks those who stayed, and those who stayed are not those who left. Its value lies in supplying the comparison group that the fifth step needs, not in explaining a departure that has already happened.
• Predicting who may leave. It asks who might leave, while root cause analysis asks what drove out those who left. The first is a forward estimate that can be right or wrong; the second traces back through events that happened. A prediction produces a list of names that needs care in how it is circulated, whereas root cause analysis produces a factor to be changed.

Cases that invalidate attrition root cause analysis unless they are separated

• Termination by the organisation. It is not leaving by the employee’s decision. Mixing it with resignations mixes the organisation’s decision with individual decisions, and the analysis ends up explaining the organisation’s policy by its employees’ reasons.
• A fixed term contract ending when its term expires. It is an end built into the relationship from the start, so treating it as a case that needs a root cause produces a cause for an event that is not a problem. The rules on such contracts have their own sources, such as our guide to fixed term and indefinite contracts.
• Seasonality. Departures that recur at the same time each year need comparing with the previous year, not with the previous quarter; otherwise a stable pattern is taken for a new decline.
• The time between cause and departure. The event that produced the decision may precede the resignation by months, so a window that stops at the last month captures the effect of the decision and misses its cause. In the sources we reviewed, we found no published period on which to base the length of that window, so it is set from the sequence of the cases themselves.
• Looking at leavers alone. It can produce a cause that looks solid: a factor present in everyone who left, which then turns out to be present among those who stayed as well. Comparison with those who stayed, the group a retention rate counts, is not a refinement added at the end. It is what separates a factor that accompanied departure from a factor that produced it.

The data behind attrition root cause analysis and its limits

What is collected for the analysis is employee data, and it is personal data under the Personal Data Protection Law (نظام حماية البيانات الشخصية), issued by Royal Decree M/19 of 9/2/1443H and amended by Royal Decree M/148 of 5/9/1444H. Article 1(4) of the Personal Data Protection Law defines personal data as any data, whatever its source or form, that identifies an individual specifically or makes it possible to identify them, directly or indirectly.

The practical effect on the analysis is that results are presented as a pattern at the level of a job or a department, not as a comparative file on named individuals. Showing them to someone who does not need them for their work widens the circle of access with no purpose to justify it.

The steps of the analysis are themselves processing of this data. Article 1(5) of the Personal Data Protection Law includes collection, keeping, retrieval, use, disclosure and sharing in the meaning of processing, so it cannot be said that the data already existed and was merely looked at again. Looking again is use, and gathering what was scattered into a single picture amounts to disclosure and sharing for whoever receives it.

The definition of attrition root cause analysis does not summarise the further duties that the Personal Data Protection Law places on an employer. Those duties have their own sources, and leaving them out of the definition does not mean they do not apply.

A methodological limit sits alongside the legal one. What is gathered about a person from someone else remains the opinion of whoever said it, not a fact about that person. Asking colleagues why a colleague left produces impressions that are recorded and then, months later, taken as facts. This is the route by which the second error described above gets in, because an impression that is repeated gains plausibility from repetition alone.

Where attrition root cause analysis should arrive

An analysis that ends at “weak loyalty” or “the labour market” has not arrived anywhere, because it has handed the result to a factor outside the organisation’s control. An analysis that ends at a pay gap in a particular job, at a manager around whom cases cluster, or at a promotion path that is blocked after two years ends at a decision that can be taken and whose effect can be measured afterwards.

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