Discover how to easily structure and breakdown your problem to find potential root causes using sequential analysis explained by B. Robert Holland. Learn to avoid the pitfalls people fall into when given a complex problem.
When first given a problem, there are usually two reactions.
For the experienced problem solvers they are likely to assume a preconceived structure since they may have worked on a similar problem. This habit is something we all have the tendency of doing, but it violates the fact based hypothesis approach we should follow.
For the inexperienced problem solvers they are likely to be intimidated by the intricacy or abstractness of problems. What is important here is to take the bigger problem and dissect it into a more manageable problem and repeat the process. As Ethan Rasiel suggests in his book The McKinsey Way, no business problem is immune to the power of fact-based analysis.
We will use Barbra Minto's version of Robert Holland's Sequential Analysis to break down the problem so we have more clarity:
What is the problem?
Where does it lie?
Why does the problem exist?
What could we do about it?
What should we do about it?
What is the problem?
When defining a problem we want to be a precise as possible, preferably with a metric that is measurable and can be used for comparison. Gathering data on performance metrics and intermediate performance metrics may help. This process would utilize techniques used for descriptive analytics to identify any existing trends or patterns. Visualizations using bar graph, pie charts, word cloud, bubble graphs, and line graphs.
Where does the problem lie?
When finding the problematic component of an operation, we need to first get a picture of the whole process. Diagramming the chain of operations will be extremely helpful in identifying the check points to inspect, thereby potentially pin pointing the general stage at which the problem arise. Don't stop until:
You have identified all parts of the operation so as to not investigate problem at the wrong stage
You have correctly ordered them in sequential order to avoid confusion in potential causal effects
Identify the inputs and outputs as they say in machine learning, "garbage in garbage out"
If we are provided with good data, we can apply methods of diagnostic analytics such as correlations and causation analysis. If we want to be even more exact, we may pin point the issue using anomaly or outlier analysis.
Why does the problem exist?
Depending on the stage at which operations have issues, we should be able to come up with some directions of why the component is not operating at optimum. In the case where the issue is still too complex, we can take a deeper dive and investigate how each element relate to each other to identify an inefficiency. There will always be many options to investigate but not enough time to go through them all, so we do have to be selective.
What could we do about it?
Once we have diagnosed the root cause, we should also have a good idea what options we have to improve it. When choosing options we must also consider the constraints we are working with. If we are being really technical we are basically formulating an optimization problem as such:
Of course most cases we don't need to go that technical, but we do need to assess which option would maximize KPI, most likely to succeed under feasibility conditions.
What should we do about it?
With a chosen alternative it's important that we visit the following considerations to assess the risk of implementation:
The gap between the current operations and the desirable operations
The structure of the situation which caused the gap
Structure of the underlying process to close the gap
Alternative ways the structure could be changed to customize
Changes required to accommodate the new operation strategy
This section ties in with prescriptive analytics in which simulations would be helpful.
Comments