Key takeaways
- Root cause analysis (RCA) distinguishes the visible problem and its symptoms from the underlying cause, and establishes a causal link between cause and problem.
- An RCA has four steps: identify the problem, determine the timeline, separate the main cause from other causal factors and establish the causal link.
- Steering on one KPI, such as the number of young people helped into work, leads to relapse of groups that cannot stay in work sustainably.
- “The target group” does not exist: look for relevant subgroups and intervene per subgroup with small PDCA cycles that demonstrably work.
- Targets cannot only be imposed top-down; they must partly be derived from the data.
What is root cause analysis?
Root cause analysis (RCA) is a method for tracing a visible problem back, step by step, to its underlying cause, so that you solve the cause instead of fighting the symptoms.
The first part of the NPRZ case described how the partners in the National Programme Rotterdam Zuid guide people back towards work. But it is not only about getting people into work: it is about keeping them in work. That is where data-driven working and root cause analysis come in.
The problem with one KPI
If you use “the number of young people helped into work” as the only KPI (the Plan phase of the PDCA cycle) and aim all your actions at it (Do), a problem arises. You then also try to help people with addiction problems or without a fixed address into a job. But people with an addiction, a complex home situation or no roof over their head will drop out. You simply know that this group will be back on benefits in no time.
The thicker the flow of people falling back into benefits, the smaller the population for whom you really make a net difference. Thanks to data and the presentation of the real, integral figures, those involved can understand how all movements towards work or school and all relapse mechanisms fit together. The measurements show exactly which people from which categories have fallen back (Check). Once you see that, you almost have the moral duty to adjust your plans for these groups (Act). In short: if you want to help people into sustainable work, you must solve their underlying problems. You carry out a root cause analysis.

The four steps of root cause analysis
- Identify the problem you see (this is not the underlying cause) and describe it as well as possible.
- Determine the timeline: start from the normal situation and stop at the moment the problem manifested itself.
- Distinguish as sharply as possible between the main cause and other possible causal factors.
- Establish a causal link between the main cause and the problem through analysis.
In step 3 you draw up a tree diagram in which you unravel the observed problem into symptoms, the possible root causes and the finally established underlying cause.
Fine plans and unrealistic targets demotivate
Mark de Kort, adviser at the NPRZ: “You can make fine plans in advance, for example by agreeing to help 3,000 people into work. But what if 2,000 of them cannot do that at all, and you know beforehand that they will come straight back onto benefits? That not only costs society a lot of money; people also become demotivated. They thought: I am getting better, I am going to get work. The consultants who guide them are fed up too, because they have the thankless task of motivating people again.”
Choose tailor-made interventions
Instead of one unattainable goal, strive for many smaller attainable ones (SMART goals). If you steer on one-sided objectives without root cause analysis, you do not do justice to complex reality. It is far better to find out which relevant subgroups the population consists of and to intervene per subgroup as needed.
An illustrative example: instead of half helping or not helping 1,000 people, you can better really help 200 people in way X, 50 in way Y and 100 in way Z. Instead of the empty promise to guide 1,000 people sustainably into work, first work data-driven: see what the data tells and then what you can really deliver, taking into account all causes, movements and relapse mechanisms per subgroup. You see that approach X works for group A, while another group is served far better by first helping them get rid of their addiction. This prevents people from being “pumped around” in the existing system.
Massive policy versus tailor-made based on RCA
A precision strike instead of a scattergun approach.
Massive policy
Tailor-made with RCA
Goal
One large target
Many small attainable goals
Population
“The target group”
Relevant subgroups
Targets
Imposed top-down
Partly derived from data
Intervention
The same for everyone
Per subgroup, on the cause
Effect
Relapse and demotivation
Sustainable results
Massive policy does not work: look for the root causes
The art is not to think from one massive policy, but to work towards a network of diverse policy lines in order to become individually relevant. Data-driven working makes this paradigm shift possible, because the technology enables ever more fine-grained policy. You get ever smaller PDCA cycles, but ones that you know work and contribute structurally to the total effect. Look per target group for a small PDCA that works. Which kind of work can each subgroup sustain, and why? What are the root causes of the problems? What is really needed to prevent whole subgroups from falling straight back?
Set targets you can really achieve
De Kort: “Targets cannot only be imposed from above. They must partly be derived from the data. We make this transparent and discussable. Client consultants increasingly tell coordinators: this target is a clear translation of political ambitions, but in practice it is unattainable. These people still have debts or receive supported care, so they cannot work sustainably. So we first check, on the basis of data, whether other problems are at play. You do not ignore that; you solve it first.” Where the original plan in a district is to help young people into work, you discover that different groups need different interventions, with intensive personal guidance where necessary.
The balance after two years
In Rotterdam a start has been made with continuous improvement based on data. Test scores at primary schools are rising and pupils in Zuid are catching up with pupils elsewhere in Rotterdam and in the four largest cities. More pupils choose training with a prospect of a job in care and technology. The percentage of benefit recipients has not yet fallen enough, but Zuid is catching up with the G4 and in fact achieves the strongest decline of all. Property values in Zuid are rising, a start has been made on the run-down private housing stock, and house hunters from the north bank and beyond increasingly consider a home in Zuid.
How Diamond Decisions uses root cause analysis
Diamond Decisions does not stop at the indicator that turns red. In business modeling we map cause and effect, with data analytics we test which cause really explains the problem per subgroup, and we help decision makers to set targets that are derived from the data.
Conclusion
Root cause analysis replaces one large, empty promise by many small goals that really work. The developments in Rotterdam Zuid are hopeful and form the proof for residents and partners that their efforts have effect, not least thanks to an approach that keeps asking for the cause behind the figures.
Talk to the author
Mark de KortDiamond Decisions
Discuss your question with Mark
Tell us where decisions in your teams get stuck today. In a no-obligation conversation we explore how everyone can make better decisions with data every day.
Schedule an introduction →Step 4 of our approach
Decision Makers
From data to better decisions every day, in every team.
Explore Decision Makers →Frequently asked questions
01
What is root cause analysis?
Root cause analysis (RCA) is a method for tracing a visible problem back to its underlying cause, so that you solve the cause rather than the symptoms. It ends with a demonstrated causal link between cause and problem.
02
What are the four steps of a root cause analysis?
Identify and describe the problem, determine the timeline from the normal situation to the moment the problem appeared, distinguish the main cause from other causal factors (with a tree diagram) and establish the causal link through analysis.
03
What is the difference between a symptom and a root cause?
A symptom is what you observe, such as people falling back into benefits. The root cause is the underlying reason, such as debts or addiction. Removing a symptom brings the problem back; removing the root cause prevents it.
04
How does an SME apply root cause analysis?
Take one recurring problem, for example late deliveries or returns. Describe it, draw the timeline, list possible causes in a tree diagram and test with your own data which cause really explains it, per customer or product group.
Further reading
About the author
About Diamond Decisions
Diamond Decisions helps organizations become better by enabling better decisions. We don’t start with data, dashboards or AI, but with the decision that needs to improve, and work back from there to KPIs, data and technology. We do that in four steps: inspiration and alignment, business modeling, analytics and data platform and data-driven decision makers. That way BI and AI don’t remain an IT project, but deliver value every day.


