In my previous life I was an analytical chemist. The company I worked for sold measurements. Environmental engineers hired us to measure something. For example, we worked on the Woburn, MA superfund site. This site, which was the subject of the movie, “A Civil Action”, contained several wells that were contaminated with TCE (tricholorethene), a known carcinogen. Engineers would collect well water samples and we would test them for a variety of volatile organic compounds including TCE.
In the lab, we had a clearly defined process which we implemented before we began measuring ANY client samples. Our analyst would spike a blank water sample with a know concentration of the compounds we were interested in quantifying. These were called QC (quality control) samples.
Our laboratory collected historical data on all the QC samples which were analyzed. This would allow us to predict or determine the accuracy of the measurements we were making. In other words, we KNEW how CONFIDENT would could be in the measurements we were providing our clients. This was really important because our product was the measurement. Our clients would then use those measurements to make decisions.
Why do I say all this? Well, if you’re measuring things in your business (you are measuring things right?) then you should also have some information regarding how confident you are about those measurements. Just measuring things is not enough. You need to understand WHY you’re measuring things and WHAT those measurements mean.
In mathematical terms, the measure of the variability of measurement is called the standard deviation. If you’re familiar with Excel, it’s pretty easy to calculate. The standard deviation tells you how precise or spread out your measurements are which also tells you something about the PROCESS that was used to generate that measurement.
Let me give you a real world example. Say you’re measuring the number of closed deals your company is generating and you’ve measured this for a year. Let’s also say the average is 100 deals per month.
How confident are you that you can predict how many deals you’ll close next month? By calculating the standard deviation you can determine a confidence limit. To do this, calculate the standard deviation and multiply that times 3. Add this number to the mean of your monthly sales and you have your upper control limit. Subtract this number from the mean and you have the lower control limit. You can now be 99.9% sure that the number of closed deals will fall within this range.
You may be shocked to find out that the range between the lower control limit and the upper control limit is quite large. For example, if the range is 20 – 130, your process for generating sales may have some problems. Your job is to figure out why and this is what metrics can reveal.