Understanding Metrics problems- 1- Metric Type
If you are in a role that requires monitoring employee performance or team KPIs, then this article is certainly important for you.
?
When you monitor performance using specific metrics, it's essential to keep in mind several important considerations. These factors could potentially lead to problems that you might encounter while defining or using the metrics.
?
In the upcoming articles, I will discuss these points in detail, starting with the first one: the type of metric you are dealing with.
?
There are a numerous types of metrics, and it is crucial to have a good understanding of their distinctions to handle them effectively. While there are many varieties, I will specifically address those that often result in confusion.
?
1- The Distinction between Linear and Logarithmic Metrics
?
Picture this scenario: you have a Key Performance Indicator (KPI) and a team of multiple members. One member achieves a score of 7 out of 10, while another is 8 out of 10 on the same metric. At first glance, it might seem that the two scores are closely aligned, and the disparity between them isn't significant.
However, this perception can sometimes be misleading. To begin with, you must know the nature of the metric. If the metric follows a linear scale, your initial assessment holds some validity. Nonetheless, if the metric adheres to an exponential or logarithmic scale, the entire interpretation shifts.
Let's return to the team illustration. If the metric employs a logarithmic scale (log base 10), it implies that the individual who scored 8 is actually 10 times better than the one who scored 7. In this context, if someone achieves a score of 9, it signifies that they are ten times better than the person who scored 8, a hundred times better than the one who scored 7, and a thousand times better than the individual who scored 6.
Before we proceed, are there practical examples of this principle? Absolutely, one notable one is the Richter scale, although I will talk about it in more details later.
?
2- The Significance of Numerical Values
In our previous point, we explored the considerable difference between an 8 and a 7 on a logarithmic scale. However, even when dealing with a linear metric, not all numbers within the range of 1 to 10 hold equal importance.
Consider the Net Promoter Score (NPS) as an example. This scale spans from 1 to 10 and tells the likelihood of a customer recommending your product or service to others. While you might initially think of 6 or above as good indicator and anything 5 or below as bad, the NPS scale operates differently. It assigns 1 point to scores of 9 or 10, zero points to 8 or 7, and subtracts points for scores of 6 or lower.
领英推荐
This distinctive scoring approach underscores the heightened significance of extremely satisfied customers. Another instance is when making an online purchase. If a product receives less than 4 stars in reviews, you might begin to doubt its quality, even if it isn't anywhere close to the middle number ( .5 star rating). The intended purpose of the metric alters how you interpret it.
3- The Concept of True Zero
Certain metrics function within specific ranges, such as temperature. When the temperature registers at 0 degrees Celsius, it doesn't signify a complete absence of heat (I'm not referring to absolute zero here). Instead, it denotes that the individual who established the metric designated this point as the baseline.
Issues arise when treating such metrics as absolutes. For example, if you state that the current temperature is 0 and then assert that it becomes 3 times colder, the value remains fixed at 0 due to dividing zero by 3. This presents both an illogical and meaningless scenario.
To further illustrate the concept, let's consider a more complex scenario. If the temperature is 10 degrees and we aim to determine how hot it would be if it were twice as hot, our intuition might lead us to believe it would be 20 degrees. However, this intuition is fundamentally flawed.
In these cases, the numerical values signify intervals rather than authentic values that can be manipulated in the same way as measurements involving a true zero.
?
4- Deciphering the Significance of Numbers
Let's revisit the example of the Richter scale. A seismic event with a magnitude of 2 corresponds to a shaking amplitude equivalent to 56 kilos of explosives, while an event with a magnitude of 3 corresponds to 1800 kilos, event with magnitude of 4 is 56,000 kilos, and of 5 is 1,800,000 kilos.
But why don't we see values like 56, 560, or 5600, considering the scale increases by a factor of 10 for each increment?
The reason lies in the absence of a specific definition of what exactly is being measured. In this case, the scale quantifies the amplitude of oscillation observed on a seismograph's trace, rather than the actual explosive power. If you're uncertain about the precise nature of the metric being used, the numerical value may lack significant meaning, or at the very least, you won't fully grasp all potential implications.
These are some critical considerations when dealing with metrics. The forthcoming article will address yet another type of challenge that can emerge in the context of metrics. Should you have any inquiries, please don't hesitate to ask in the comments section