What makes an effective Key Performance Indicator (KPI)? How do you design your metrics to help your company achieve its objectives?
First off, a good KPI will actually help you make a decision. The information should clearly tell you where you’re at and give you something actionable to do.
Secondly, a useful KPI will typically be a ratio—one thing divided by another. For example, if your goal is to decrease your employee attrition rate, simply knowing the number of people who leave per month isn’t meaningful. If ten people left last year and 15 left this year, that sounds like things are moving in the wrong direction. But what if your total number of employees has increased dramatically in that time? Let’s say you had 100 total employees last year, but 200 this year? As a ratio of the total number of employees, your attrition rate has actually improved. That’s why it’s important to look at the ratio.
The final key point is that this ratio needs to be observed appropriately in time.
For example, imagine if your company has the goal of recruiting and hiring more women for a particular role. Here is a hypothetical look at the numbers for this role in the past five years:
Year Employees in role Female employees in role % of female employees
2013 16 0 0
2014 24 1 4.2
2015 30 2 6.6
2016 35 3 8.6
2017 40 5 12.5
If you only observe the data for 2017, it might not look that impressive. Only 12.5 percent of people in that role are female. However, if you look over time, that’s a substantial jump from 2014, when only 4 percent were women. So that’s a 300% improvement!
Taking it a step further, compare each year’s percentage of new hires for the role that are women.
Year Employees hired Female employees hired % of new hires that are female
2013 6 0 0
2014 8 1 12.5
2015 6 1 16.7
2016 5 1 20
2017 5 2 40
Even looking at the difference between 2016 to 2017, it might not look that impressive. Okay, so you hired two women—which was one more than last year. That doesn’t seem to even make a dent in the number in the grand scheme of things. Because you’ve had so many men in this role for so long, you still have a low overall ratio of women in the role. But if you look at the numbers a different way, two out of five total hires were female, which is 40 percent. So you’re currently hiring women 40 percent of the time, whereas over time you’ve hired at 12.5 percent. Year over year, this is clearly an improvement. You’re doing much better now in terms of gender diversity, which is really all that matters because it’s all you can focus on and improve.
Going back to the first point, looking at ratios in specific periods of time is what makes your KPI actionable. You can use this information to make decisions and see if you’re achieving your performance goals. In the above example: Is your company meaningfully improving its ability to hire or recruit women into this role? In this case, the answer is yes. There has been a dramatic improvement each year. But you wouldn’t see this if you just looked at all of the data. Looking at it over time in cohorts tells you what’s really happening and makes it a more effective KPI.
If you look at your data year by year, or even quarter by quarter, then you can clearly see if something has improved or gotten worse. In contrast, staring at a cumulative view tells you nothing that is truly actionable.
You can make your KPIs actionable by making them ratios (something divided by something) and studying them in cohorts vs over all time.