I have seen a couple of examples where industries or departments show a KPI where it always hovers around 99.7% to 99.9%. Almost all the time the KPI chart will look like this:
Sure, you could zoom the chart in even more, but it still would not communicate what the difference between 99.8% and 99.9% is.
Let us take a closer look at an MTA metric around the percent of required buses and operators available in the AM peak period.(Pictured above) (Note they say "required", not "scheduled") It is a metric to ensure that the minimum/required amount of service is available. You have two main factors on why the buses would not be able to go out: No operable bus, or no driver to drive the bus. From the graph above, we can see that the previous 12 month average is 99.8%... Impressive!
Displaying the data this way is great to show the awesome performance, but it does a poor job of calling out how to make the process even better.
So what is the difference between a process that is 99.8% vs 99.9%? The classic example is the up time your electricity is on. In the United States, we expect it operating 99.9% of the time while some countries are grateful they can have it on for a couple hours a day.
- 99.38%: Equivalent to 60 minutes a week of no electricity (Hope you save your work often!
-99.87%: Equivalent to 13 minutes a week of no electricity (Not bad, but I cant imagine knowing that we would be on backup power 15 mins every week)
- 99.977%: Equivalent to 10 mins a month of no electricity.(This is probably normal, summer storm knocks power out for 2 hours on a mid august day, that is still a 99.977% uptime for the year.
- 99.99966%: Equivalent to less than 2 minutes a year of no electricity.
So we can easily see that there is no way I am accept a 99.4% success rate of electricity transmission, let alone 99.87%. I want to see MULTIPLE 9's.
ANSWER: USE PROCESS SIGMA of 1 through 6
I will let the pros explain what Process Sigma is:
When we look at the Bus metric and see the capability of dispatching the required buses continuously approaches 99+%, we should convert that to a Process Sigma rating of 1 through 6. Here is the conversion table at a high level:
Now we can have a much more dynamic looking graph as well as easily see a difference when we are talking about 99.87% (4.5) vs 99.997% (5.5).
Here is the same Bus Chart converted to Process Sigma numbers: I also like to tell my audience that if you see a process that is 1.0 sigma values different, then THAT IS A HUGE DIFFERENCE.
****Yes, of course your can keep zooming into the original graph to get the same shape as the Process Sigma, but the shape is not the point, The point is to simplify the true capability of the process and work towards improving towards the next Sigma Level. Seeing something that is 3.7 vs 4.5 is much easier to grasp than something that is 0.0002 different than the other.
To determine your performance, you need to set a goal as to what the process capability should be. Is 4 Sigma our goal or should we strive for 5 Sigma. Remember, the difference between the two is so small on an absolute scale (2.7^-5) yet from a large operation you are talking about an extra 6,000 defects per million.