The basic dogma of KPI (Key Perfomance Indicators), the core knowledge of OE, the prayer of the Operational Excellence Manager: you cannot improve what you don’t measure.
But if it was that simple, we would all become incredibly fit people as soon as we started wearing a fitness tracker, right? From personal experience I can tell you that’s not what happens, at least not all the time. While wearing the tracker, I was no longer moving because I liked moving (which I do!), I was hitting that step count and exercise minute goal. It felt like work. And worst of all, I didn’t even notice the change in my attitude towards moving, I only noticed that I was less and less interested in my tracked results. At one point, my wrist started to hurt for some reason and I took off my tracker for a while, and that’s when I finally understood the way tracking my movement had influenced me.
So what went wrong? I did measure what I wanted to improve, why did I not improve?
Regardless of exercise issues, there’s a lot to unpack around the way humans deal when their performance is observed or measured. We know from physics that the very act of observing something fundamentally changes how that something behaves. While people are a bit more complex than simple objects, there’s still something intriguing about this. How does the fact that we track and monitor certain things affect people and their behavior, and are there any more unintended consequences in addition to what I’ve seen in myself? And how does this relate to the way we need to define a KPI?
So let’s discuss some scenarios where KPIs failed to achieve their purpose. I’ve chosen these because I’ve seen the described dynamic play out in real life, but because I wanted to give some specifics, we can’t discuss the actual examples – instead, we’re going to talk bakeries and investigator teams. The described patterns should apply regardless of industry. There’s also some ideas for solving the described problems, but I’d love to hear your take on this as well – I certainly don’t have all the answers.
KPI Failure Mode 1: KPI is not sensitive enough to performance changes
Let’s say you run a bakery with a couple of motivated and competent bakers. They make the best cakes in town, so you must be a great bakery overlord – congrats, job well done! Let’s also say your KPI is number of cakes produced with “best in town” quality. Of course, they also need to be on time, because belated birthday cake is just sad for everyone involved. So: number of cakes, on time, and on quality target. Sounds reasonable, right? Now let’s imagine one of your bakers becomes less motivated as the others. In fact, they suddenly get kind of sloppy and start making mistakes.
You might expect to see this person’s lower performance reflected in the KPI immediately, right? But no team runs at 100% of their performance maximum in normal operation – it would make vacations and sick days devastating to your business. There is always some excess capacity around. The rest of your team is still competent and motivated, they just love making people happy with great cakes so much. And also, you as their bakery overlord are on their tail constantly about that damn KPI – in this situation, the team will tend to cover for the suddenly low-performing baker. They might rework the slacker’s cakes on the side, or they might spent more time overseeing their work, especially when they like the person or the person has a good (personal) reason for a sudden decrease in quality. People are people, and your team knows that it could happen to them as well, so they will pour in more effort in order to keep the KPI up.
In this case, your KPI will not react to a difficult problem in your team structure early enough. The KPI will only then start to drop when your motivated people get fed up with covering for the low performer, or burn out, or just plain cannot keep up. By then, you have lost more than just a bit of performance – you have lost the motivation of the team and the team coherence, or maybe burned out a couple of your loyal baker minions.
Measuring overall performance alone will not necessarily give you a clear picture of problems as they occur, because motivated organizations will cover performance drops with increased effort and energy spent – until that energy runs out. So you want to define your KPI in a way that let’s you notice when a performance drop is sneaking up on you, which means at the root of your perfomance.
Now, I am not saying to fire the slackery baker (that sounds like an awesome name for a bakery btw). In fact, it’s quite the opposite, they need your undivided attention to figure out what they need to contribute equally to team performance. But for that, you need to know there’s an issue, while your KPI-driven organization has every incentive to mask the issue for as long as possible. This is an excellent use case for combining OE with the systemic approach, especially when applied as prevention instead of in crisis.
KPI Failure Mode 2: The KPI is not objective
Alright, time for a job change! Your new task is to run a team of investigators. Your team is responsible for solving misteries, like where did you leave your car, how do we stop the climate crisis, and what’s the best curry recipe in the world (any leads on that one?). You would like them to track the progress in their current case so you know when you can assign a new case to them. The thing is, though – how would they know? You can only say you’re done finding a car once you found it. In the meantime, it’s either in the next street you search, or it’s not.
They will probably be able to give a good estimation in cases that come up often, but most of the issues will be unique and therefore impossible to estimate. This is one of the main problems with project work and KPIs.
In an organization that values KPI too much, where your team will be forced to give a number, any number, they will tell you what you want to hear – but it will be meaningless. If a KPI cannot be based on measurable input, I’d think very carefully about these risks before I implemented it. Maybe there’s an option that can be more easily assessed objectively? E.g., if you break down your investigation into smaller tasks, you can track the current stage of the investigation. Or you could track active tasks for each employee and their current status with a Kanban-Board. Agile software development uses this heavily for a reason, or so I’ve heard. I’ve never actually seen an agile software team. Either way, the board gives your team the option to share their progress and struggles without forcing them to put a meaningless number on the board. Meaningsless tasks are not just a waste of time, they will also kill any motivation towards better performance your team might otherwise have. You also lose any motivation or benefits that the team might gain from useful KPI that support them in assessing their perfomance fairly and identifying necessary improvements.
There’s more ways KPI can fail to come. Let me know if there’s something specific you’d like me to discuss in more detail as well!
Moderation note: I am disabling comments on the blog. If you wanna get in touch, please do send me a message via the social media network of your choice!