Calling my new ISP
Two days ago I tried calling my new internet provider and I was definitely not the only one with the same idea.
After a few minutes on hold I hung up and sent them an e-mail. Their customer service still hasn’t replied.
I’m considering a different internet provider now. They have no idea about my sentiment of course.
This reminded me of a story from the second world war.
During WWII the navy tried to figure out at which spots to armour their planes.
Too much armour would make the planes too heavy, so it was important to make some choices here.
In order to make a good choice they started mapping where all the bullet holes were in the planes that returned from combat.
They found that most bullet holes would be in the tips of the wings, the tailwings and the mid-section of the plane.
At first sight, those were the spots that would need reinforcement.
Adam Wald, a statistician, would disagree. Mister Wald was smart enough to ask where the missing holes went.
After a while the answer started to dawn on his teammates.
The missing bullet holes were in the other planes, the ones that didn’t return from combat!
The navy thought they were analysing where their planes would be hit most often.
Instead, they were analysing which bullet holes did not result in planes crashing down.
I don’t need to explain where they added the extra armour!
The phenomenon later became known as survivorship bias.
What you see vs. what you don’t see in your data
If you’re in customer service, especially within workforce management, you probably at some point tried to analyse what wait times would be acceptable to your customers.
It is considered good practice to plot waiting times against customer satisfaction metrics, in order to find the sweet spot.
Below plot is an example of to illustrate this.
This is a scatter plot of wait times in minutes against customer satisfaction at a doctor’s office. (read the full article here)
The takeaway from this plot could be that there’s a sweet spot between 10 and 20 minutes wait time.
This would minimise idle time for the doctor himself and not test the patients patience too much.
Also, more than 20 minutes of wait time would definitely be undesirable.
But what about the customers that avoid seeing the doctor all together, because of the anticipated wait times?
And where would these chaps go when there’s a new doctor in town that advertises zero wait times?
A study by Arise figures that 13% of respondents does not find any wait time in customer service acceptable at all.
A total of 16% already does not even bother trying to call customer service anymore:
So how much would gaining 16% market share be worth to your company?