The last blog posting on “The Characteristics of Metrics that Matter: Actionable“, focused on a key metric usability question – What good is a clearly defined metric if you cannot understand why it changed and therefore what action to take? In this blog, I focus on the time factor. How can you take action if you don’t understand what happened quickly enough? Do you have historical knowledge to know the effect of your actions? Those two questions are the basis for the Characteristics of Metrics that Matter: Timely and Time-trackable.
Warning: You Are About to Go Critical
To ensure timely response, a metric must be updated frequently enough to provide the signal to act quickly enough. Similarly, your decision to take action and the action itself must not take too long or your efforts become diluted or useless. Sadly, real-time data collection, metric calculation, reporting, and action is not typically realistic. Real-time all the time sounds great but, from an efficiency point of view, is it worth the price or even possible? For instance, you may have some control over collecting and reporting internal customer service metrics (during and just after contact-based touch points) and thus can act more quickly. However, things like external customer surveys may have uncontrollable data receipt delays. For some metrics, you may need to set up an early warning prior to crossing a critical threshold, such as an unacceptable delay between the time a condition changes, the associated metric hits its critical threshold, and action can be taken.
Trendy Isn’t Always Disposable
A metric that matters should also be “time-trackable”, that is, recorded over time so insight can be gained as values are observed. Without a history of metric data points to analyze, you can’t easily observe common or exceptional trends or know if your responses have long-term effect. Not only can trends like seasonality be monitored but observations over time indicate if a goal or threshold is correct for your purpose. In addition, it is important to monitor over the right time frame for the right meaning. For instance, a weekly trend may be totally irrelevant (“noise”) when compared with a quarterly trend. Finally, time-trackability is a requirement for predictability since a study of the past is key to discovering ways of predicting the future. In my next blog, we’ll talk more about predictability and comparability with respect to metrics that matter.
How are you tracking your key metrics over time and what have you learned? Please let us know in a comment.