Finding the MOST Effective Solution — 2
In this part of my series on problem-solving, I explore one of my favourite problem-solving techniques (root cause analysis) in detail. You won’t get the Donald Trump references if you haven’t read the introduction to this series, so go check it out.
Correlation ≠ Causation ≠ Root Causation ft. Donald Trump
Root cause analysis focuses on problems that have already happened, unlike many other techniques; instead of solving a new problem, it is often used to prevent a problem from happening again.
The process for root cause analysis has four basic steps:
- Organise all known information about the problem/accident;
- Create a timeline starting at normal operations and ending at the problem occurring;
- Identify all the correlations and causes of the problem;
- Establish the root cause(s) which led to all the other underlying causes.
The goal of this process is to organise the steps that went wrong in regular operations to lead to the failure/problem and then work backwards to find the first point of failure (or the root cause). Fixing this root cause should permanently fix the problem and prevent all other causes from happening, which is why this process is more effective than trying to chaotically fix whatever is most obvious.
It’s easier to understand the benefits of this process with a concrete example. Consider the anguish many had after 2016’s American election, wondering what possible explanation there could have been to land Donald Trump in office. Perhaps, some might blame extreme world events that created an increased atmosphere of mistrust, whereas others might identify these as mere correlations. In comparison, you could identify Donald Trump’s campaign speeches as a factor that caused his election. Looking deeper, however, it might actually be the blunt and harsh content in his speeches that characterised his campaign, driving people to support even radical policies. In this small (and incomplete) chain of events, the root cause for Donald Trump’s election would be his use of blunt, fear-invoking media to garner support. If this had not happened, his speeches would not have been as effective and he would have been less likely to be elected.
If someone were to, for whatever reason…, go back in time and stop Donald Trump getting elected, the person would have an easier time making the content of Donald Trump’s speeches less radical than stopping them altogether. Evidently, eliminating the root cause is less obvious than eliminating a surface-level cause or correlation, yet more effective. This addresses the problem efficiently and also prevents it from reoccurring.
In a more typical example, an engineering team might be assessing the root cause of the failure of an electric car’s battery. They might determine that the battery failed because of a significant increase in discharge. This increase might be caused by other factors such as the battery’s electrolyte having chemically changed over time and could be correlated with events such as cold temperature conditions. The root cause of the chemical change in the electrolyte, however, might be inadequate maintenance due to the car’s recommended service periods being longer than those of the battery. This leads to the battery not being serviced (even after its service period) until the car is, causing damage to the electrolyte. Thus, the problem would most effectively be solved by shortening the recommended service cycle for the car so that both the car and battery receive maintenance more often. In addition, this solution would also address the problem permanently, instead of simply replacing the malfunctioning electrolyte (which would eventually deteriorate again without proper maintenance).
Challenges with Root Cause Analysis
- Often, companies cannot collect and store detailed data for all their systems for long. This means that key data is often missing when trying to identify the root cause.
- It can be challenging to establish a timeline from the point of regular operations to the point of failure to see which initial root factor caused other systems to malfunction.
- In the real world, there can be multiple root causes because of the complexity of machines and systems. Fixing one can lead to worsening the other.
- The root cause is different for multiple people. A root cause for an engineer may stop at the last point of failure in design, but it might extend to relevant maintenance procedures for a manager.
In fact, my very first time using root cause analysis did not work, as I tried to use it for the wrong type of problem; I was trying to find the root cause of someone littering at a bus stand. The major issue with using root cause analysis for this is that the problem is not a closed system. This makes it hard to establish a timeline of events from the person waiting at the bus stand regularly to the person littering. Due to this, I could not isolate causes and correlations without getting distracted. For instance, the person littered because there was no trash can close by, because they didn’t have enough time to find one, because they were in a hurry to get to work, because… Clearly, it is easy to get carried away in this case, demonstrating just one of the challenges with this problem-solving technique.
If done right, however, root cause analysis enables efficient solutions that address a problem over the long-term. That being said, this technique is not effective in open systems with many interacting variables. For these problems, it can be useful to consider a technique like means-ends analysis.
Wait… How are the AI Overlords Taking Over Again?
That’s right; Means-ends analysis is the problem-solving technique used in artificial intelligence. The specifics of its implementation is a topic in itself, especially with computer algorithms, but it is actually a simple mindset to follow for human problem-solvers.
Means-ends analysis constantly focuses on the goal in mind when trying to solve a problem. Each action taken should bring you closer to that goal by minimising the difference between where you are now and your goal. Maintaining this mindset when solving a problem allows you to analyse the effects of each action you take and ensure that your efforts bring you closer to solving the problem.
Of course, it is important to realise that when many factors influence each other, sometimes there is no perfect solution. You simply have to choose the trade-off that gets you closest to the ideal outcome. The Russian TRIZ theory, a problem-solving tool for inventors, even mentions that the most valuable problems are the ones where there is no perfect solution to optimise; an inventor must find ways to bypass this or make the tradeoff that yields the most favourable outcome. By always focusing on the goal, you can be objective and choose the most effective trade-off while trying different solutions.
Take an example from the automotive industry: increasing sales by improving the fuel efficiency of a typical family car. The manufacturer may consider reducing the weight of the car by designing only with the least amount of materials possible. Consequently, the car would be more fuel-efficient and inexpensive, however, it might also have a reduced margin of safety in case of accidents. On the other hand, the manufacturer could design an engine with increased efficiency in mind, however, this might drive up the price and reduce performance metrics. In this case, the important factor is to keep the goal of improving sales in mind, while making trade-offs that bring the car closest to this goal. This may include increasing fuel-efficiency in some aspects of the design while sacrificing it for other factors elsewhere.
As can be seen, means-ends analysis can be a useful mindset to keep in mind to acknowledge that a problem may not realistically have a perfect solution. The problem-solver must find the ideal solution as effectively as possible, without wasting any additional resources.
Key Takeaways
- The root cause isn’t always the most obvious point of failure;
- Addressing the root cause(s) fix a problem permanently;
- It takes extra effort and maintenance to perform root cause analysis;
- Means-ends analysis constantly hones in on the goal;
- And the perfect solution may be impossible. You have to find the most ideal one.
Despite the usefulness of root cause analysis and means-ends analysis, both have one common flaw; they require careful planning and consideration that takes up valuable time. For problems requiring immediate action, there are more effective techniques like lateral thinking which are the focus of the next part of this series.
