Matthew Renze

Metrics Have Consequences

How to choose effective metrics for your team, project, and organization.

Author: Matthew Renze
Posted: 2018-10-12

Metrics Have Consequences

What would happen if every doctor in the world was rated solely on the number of lives they saved? At face value, this seems like a good measure of success. A doctor with a higher life-saving ratio would be better than a doctor whose patients were constantly dying.

However, metrics have consequences. As a result of doctors being measured in this way, several unintended negative consequences would certainly emerge. For example, if a doctor's salary was tied to this metric, they would almost certainly begin gaming the system to improve their salary.

As a result, we would see doctors refusing to take on elderly or high-risk patients. We'd also see life-saving procedures performed on patients without any life-threatening illnesses. In addition, we'd see patients being kept alive well beyond the practical and ethical limits of medicine - all to improve a number.

The same is true with the metrics we choose for our teams, our projects, and our organizations. There are metrics that will drive positive outcomes and there are metrics that will lead to negative outcomes. The challenge is deciding which metrics make the most sense for you and your organization.

The following guidelines will help you to decide how to choose effective metrics to drive positive outcomes for your organization. Effective metrics are:

Purposeful

Good metrics should drive business outcomes. This starts with your organization's "why" (i.e. it's purpose for existing) and forms a hierarchy of KPIs down to your day-to-day operations. For each metric, ask yourself, what outcome is this driving, and how is it ultimately connected to our overall purpose.

Actionable

Good metrics lead to action. You must ask yourself the question, "if this metric changes, what will I do in response?" If you don't have a clear answer, it's likely a bad metric. Metrics that are not actionable are called vanity metrics . They make us feel good but they don't lead to any real action on our part.

Reliable

Good metrics are trustworthy. One of the top complaints I hear from decision makers is that they don't trust their metrics. Metrics are only as good as the data and analysis they are built on. As a result, you need to make sure your metrics are based on clean data and statistically-valid methods of data analysis.

Valuable

Good metrics provide more value than the cost to produce. If the cost to create and update a metric is greater than the value it provides your team, then it's a bad metric. However, if the insight you derive from a metric is more valuable than its cost, your organization is growing as a learning organization.

Timely

Good metrics provide you with the insight you need, when you need it. If a metric is not available when a decision needs to be made, it has failed at its job. If a metric involves a window of time that's too large or too small, it has failed as well. Metrics need to be available on time and with appropriate time scales.

Impersonal

Good metrics are not used to punish or reward individuals. Human behavior inevitably changes in response to being measured. When measured, we will "game the system" in our favor. We avoid this observer effect by using metrics to guide the improvement of systems, not to measure the performance of people.

Frictionless

Good metrics require little to no manual effort. If a metric takes significant time and energy to create and update, it is less likely to be reliably maintained. As a result, we want to make data collection and analysis as frictionless as possible; ideally, using automated data pipelines from operational systems.

Necessary

The best metric is no metric at all. If you can effectively solve a business problem without relying on a metric, you should attempt to do so. In Agile organizations, we prefer individuals and interactions over processes and tools. So, avoid relying on unnecessary metrics and trust your team to get the job done.

If you'd like to learn more about using data to drive decision-making practices, please check out my online courses on data science.

Share this article:  Share on Facebook Share on Twitter Share on LinkedIn Share on Google+