When you bundle a lot of computer wires together, it gets hard to figure out which is doing what and why. The same can happen with all of the metrics that IT collects, and it is possible that you are collecting metrics that do not really align with strategy or producing value. In an article for InformationWeek, Dawn Parzych describes how you can stay out of the large “danger zone” of metrics.
The Wild Ride
Danger comes in many forms. For instance, you should always practice weariness when creating proxy metrics, which are metrics that “get the gist” of something when it is impossible to get a direct measure. By nature, a proxy metric will never be perfect, but you had better make sure it is in the ballpark. Otherwise, proxy metrics might morph into vanity metrics, which are metrics that make things appear good but ultimately mean nothing. The point of metrics is not to trick people (especially not yourself); it is to understand efficiencies and act upon trends uncovered.
Yet another area of danger might emerge from summary metrics, about which Parzych says this:
Means and medians are easy to understand but they often don’t tell the whole story. They hide the presence of anomalies and outliers. Often we can learn more from the anomalies and outliers than we can from the average events. A common metric used to track performance of incident resolution is mean time to resolve (MTTR). This doesn’t show whether all incidents are resolved within the same timeframe, or if some incidents take a few minutes while others take much longer. To improve this metric there needs to be an understanding of the distribution of measurements.
Parzych then continues to make these recommendations for smart metric use:
- Align your metric with an organizational goal.
- Track actionable metrics.
- Look at the big picture; do not fixate on one metric even if it is a good metric.
- Metrics should lead to collective growth, as opposed to competition between teams.
You can view the original article here: https://www.informationweek.com/big-data/big-data-analytics/avoid-the-danger-zone-of-metrics/a/d-id/1329736