Mapping Reporting and Operational Metrics with Root Cause Trees
In my last post, I wrote about Reporting vs. Operational Metrics. Reporting Metrics measure business processes as trends over time. Operational Metrics are used at a much lower level to understand processes at points-in-time or on shorter time horizons. I pointed out the relationship between the two categories: When there is variance in a Reporting Metric, the analysis of that variance always takes place using Operational Metrics.
I also wrote about a common leadership trap that one can fall into when Operational Metrics are mistaken for Reporting Metrics – the resulting behavioral spiral which leads to poor and short-sighted decision making, micro-management of front-line managers, and distraction from strategic thinking.
In this post, I will lay out at a high level, the framework I’ve coached functional teams through to organize metrics for their areas based on root cause trees.
This framework is designed to combat a few common problems organizations face when trying to make decisions based on metrics:
- Distinguishing between Operational and Reporting Metrics
- Knowing what is important in the vast sea of metrics being reported
- Difficulty understanding how metrics relate to each other
- Knowing how to analyze a metric that is trending poorly
The first step to building out a root cause tree of metrics is to gather up the right people for the discussion. Generally, I’ve had the most success with a group that includes one or two senior-level leaders, a few mid-level managers, and a select few people whose job it is to do analytics or author analytical reports. The analysts/report writers are critical because they know the data and systems, and they have experience using them to explain why something is happening.
Take this group through the following four step facilitation framework:
1. Bookend the process or function.
For example, we might be working with a process which starts with orders taken in, and ends in a customer receiving their order. Whatever the process or function is, establish a start point and a stop point that contains your thinking and prevents unrelated processes or functions coming in.
2. Establish top-level metrics.
With the group, talk about what are the highest level measures for the process or function. These measures shouldn’t roll up into each other – if they do then only the highest level of the metric is really top-level. Additionally, there shouldn’t be more than three to five top-level metrics, unless you are dealing with a major business function (e.g. Global Supply Chain). Even then, I’d stay close to five metrics if at all possible.
One other tip: Top-level metrics should directly relate to the customer in some way. If they do not, it may be time to question the actual importance of the metric.
3. Burrow into the top-level metrics.
The goal now is to understand which metrics strongly relate to the top-level metrics. I do this by picking one metric and ask a simple question: “When this metric starts to trend poorly, what do you look at to understand why?” If the response back is a comprehensive list of every metric the company measures, insert a bit of perspective along the lines of “Alright I get it, but 80% of the time, if that metric goes south, where do you look to understand why?” By doing this, you should be able to get a manageable list.
4. Rinse and repeat.
You now have a list of top-level metrics and mid-level metrics. Keep burrowing down into those mid-level metrics using the same question for each one: “When this metric starts to trend poorly, what do you look at to understand why?” Repeat the process for as many metrics levels as makes sense – I normally stop at four or five levels. You should end up with something that looks similar to the diagram below for each top-level metric (A in the diagram):
This metrics tree is powerful in at least five ways. First, you now have a fantastic way to differentiate Reporting Metrics and Operational Metrics. They naturally fall out of the tiers of the chart. If you are looking at a function’s metrics, the top level is probably reported to executives, the second level to functional managers, and the third and fourth levels to front-line managers. Anything else is probably only used for unusual levels of analysis or variance explanations.
Secondly, you have great context for your metrics. Each metric on the tree relates to (and is diagnostic of) the one above it. If there is a problem with metric “A” above, 80% of the time you will find the reason in the “B” metrics. Have a problem with B? Check out where it’s linked in C and you can diagnose why, and so on down the causal relationships.
This chart makes it abundantly clear which metrics will end up mattering most to company goals. Are you measuring (or even reporting) metrics that didn’t make this chart? This is a great opportunity to ask why – perhaps you missed an important metric when generating the root cause tree. Or perhaps you can eliminate a few less-than-useful metrics. I’ve been amazed at how many metrics some organizations can simply stop tracking because they no longer matter (if they ever did at all).
If you are using metrics for performance goals, this chart can help make sure that your incentives are correctly placed (i.e. they support company goals) as well as help you explain to your employees why the metrics matter to company goals.
Finally, there are massive reporting implications. Your reports can become more focused, with truly valuable drill-downs if they are based on this tree. In a reporting dashboard, for example, any metric with a variance could be clicked on, bringing up the diagnostic metrics one level underneath (click on A to bring up the B metrics… click on the problem B metric to bring up the linked C metrics…).
For more on the reporting implications, take a look at part three where I address metrics governance and review. Read the conclusion of this series here for ideas on how to establish a culture that will help steer your organization clear of common leadership traps.
Please let me know what your experiences and thoughts are on the subject of metrics in the comments below!