In business, we frequently say, “you can’t control what you can’t measure;” or, better yet, “you get what you measure.” We thus invest in sophisticated (read expensive) business intelligence (BI) programs run by professional teams of analysts. These BI programs frequently involve the purchase and installation of very high-end enterprise tools, supported by large databases that can collect and warehouse just about any piece of data from the business. But after all of that time and money, most BI programs don’t succeed. Why not? Because most programs focus on the technology.
The most important aspect of a BI program isn’t the technology
(that’s just the time-consuming part);
it’s actually figuring out what to measure.
In this article, I propose a framework that enables:
- understanding and categorizing five distinct types of metrics; and
- developing a hierarchy of metrics that supports achieving an outcome.
Ultimately, not all situations require all five types of metrics; however, the more complex the operation, the more likely that one or more metrics in each category is necessary with a linear dependence that looks like this (click on image for details):
These are the kings and queens of metrics; they’re the ones we hear about all of the time and measure our goals — the desired outcomes of our efforts. In the business world, examples are annual revenue, customer retention rate, and employee turnover. In our personal lives, outcome metrics include weight, body fat percent, and annual income. These metrics, by their definition, are very hard to move and thus change very slowly; more importantly, they describe the outcome only after it’s been achieved (i.e., they lag the outcome). These tend to be the easiest metrics to measure and are thus the ones we frequently talk about; strangely enough, since they’re the hardest to influence, it’s very difficult to see the impact we’re having on them on a day-to-day basis.
These metrics are predictors of the positive outcomes we desire; that is, moving these metrics in the right direction leads to our desired outcomes. They’re a bit faster-moving and are a bit easier to influence. Examples from the business world include the customer satisfaction score an agent receives after an interaction with a customer or the amount of time customers spend waiting to talk to someone. Examples from our personal lives include number of calories consumed, minutes of exercise, and number of hours worked in a day. Optimizing these metrics is a predictor of a positive outcome and are thus are leading to the outcome. These metrics typically take a backseat to outcome metrics and thus receive much less focus despite the fact they are much easier to influence.
These are the most-often-missed but arguably most-important metrics: those that actually drive behavioral change. After all, if a metric doesn’t result in someone actually doing something (or doing it differently), how will change and optimization occur? Constantly optimizing behaviors is what you ultimately need to do. Examples of these metrics from the business world include number of training modules completed and number of knowledge-base articles created (i.e., things that’ll help an employee improve his or her predictive metrics over time). Examples from our personal lives include eating out no more than once a week and reading a chapter of a book nightly. These metrics are the ones that drive habits that move predictive metrics that lead to desired outcomes. For more on the importance of habits, read Happenstance & Habits: Dual Factors Shaping Our Lives.
These are the metrics we use to operate our lives. These real-time metrics help us to understand context. Examples from the business world include number of customers in queue and number of agents currently available to take calls. Examples from our personal lives include the current temperature, a car’s speedometer, and even the current time displayed on a watch. Operating metrics are used to provide us context about our situation so we can choose the most appropriate behavior. BI dashboards frequently focus primarily on operating metrics (e.g., consider the primary gauges in the dashboard of your car).
These metrics are a bit different than the others because, in reality, they’re pseudo-metrics. That is, these metrics are “forecast” versions of the first four types. They predict what circumstances will be like in the future: what the temperature will be tomorrow, how many customers we expect to visit our website next month, or how many agents we’ll need on Valentine’s day to keep up with demand. Their primary use is for planning purposes: without an ability to forecast the future, it’s very difficult to make preparations that’ll ensure optimal conditions when that time arrives.
This post was originally published at bassamsalem.com
Deming’s challenge, i.e. to effectively manage something, it must be measured, certainly holds true, in multiple areas of business and in life. From a business decision-making actionability perspective, for me there is a metric, or more properly a framework, that enables an enterprise to get the best out of elements 1 through 4, and a bit of 5. It is customer and employee advocacy behavior measurment: http://customerthink.com/is-there-a-single-most-actionable-contemporary-and-real-world-metric-for-managing-optimizing-and-leveraging-customer-experience-and-behavior/ When this blog appeared, there was good dialogue with colleagues on the subject.
FYI, I’ve had a fairly well-read recent blog on the related issue of metric application: http://customerthink.com/comcasts-nps-gamble-can-the-metric-help-fix-the-customer-experience-culture/
Note that Deming also recognized the existence of many important things, that must be managed, cannot be measured. Both those points are important. First, an organization can’t measure everything of importance to management. And second, the organization must still manage those important things – – because they matter too.