Metrics -- the Achilles' heel of tech support

Support-call statistics can be very misleading -- all the more reason to look beyond the numbers

When you call customer service or technical support, you often hear a message that says your call is being recorded to improve your support experience. You probably don't pay much attention to this message, but its simple fact, and the system behind it, may be part of the reason Gripe Line receives so many unhappy stories about phone support.

Your call -- and nearly every other call going on around it at the support center -- is being measured, with the goal of making the support center run like a well-oiled machine. Unfortunately, once people start taking measurements, they have a tendency to pay more attention to the numbers than they do to what the numbers are trying to assess, especially hard-to-quantify concepts like customer satisfaction.

[ For a look at where tech support is going, read Christina Tynan-Wood's "The (better) future of tech support." | Frustrated by tech support? Get answers in InfoWorld's Gripe Line newsletter. ]

For example, numbers that tell you how long each technician takes to resolve issues can lead one to conclude one tech is sharper because he gets the job done faster. But it could just as easily be the case that he is better at getting customers to give up and solve the problem themselves. Both result in a quick resolution, but one leaves the customer sure he wants to do business with another vendor and the other leaves the customer satisfied and back to work.

Off the Record submissions

"A lot of key performance indicators you see today in an organization are there simply because they could be measured," says Anna Convery, CMO of ClickFox. "Average handle time is a good example. But this has manifested into a big double-edge sword. Companies should be focused on 'appropriate' handle time."

Of course, that's harder to measure.

Call escalation is also easy to measure, but it requires intelligent interpretation. For example, a first-tier technician might be quick at recognizing when he has an intelligent IT pro on the line who would be best served by being sent directly to the most advanced support in the company. However, he might be penalized by a system that assumes he's passing the buck instead of solving the problem -- simply because, when looking at metrics, it's hard to tell the difference.

The problem isn't really with the fact of the metrics. It's with the way they are interpreted. Rackspace, a company that prides itself on what it calls "fanatical support," uses metrics but weighs them carefully.

"We don't view how much time a tech spends on the phone as more important than customer resolution," explains Max Thoene, vice president of fanatical support at Rackspace. "If one of our support reps takes eight hours to answer one ticket but that customer's issue was resolved and the customer is happy, that was a successful day. We would say to the technician, 'Man you had the tenacity to stick with it!' And hopefully the customer is very satisfied."

It would be a pipe dream to think that every call could go that way and the company would stay in business. People are expensive. A call that takes eight hours to resolve is a problem. But I think I read somewhere recently that a problem is just an opportunity in work clothes -- and that's how Thoene sees it.

"We would look at that call and try to learn from it. Problems like that are great learning opportunities," he says. "It's frustrating to be faced with a difficult problem. But in resolving it, we all learn. And when we do find a solution, everyone is giving each other high fives and feeling good."

Maybe tech support and customer service centers would benefit from keeping in mind something that a wise man (namely Mark Twain) pointed out quite a long time ago: "There are three kinds of lies: Lies, damn lies, and statistics."

Got gripes? Send them to

This story, "Metrics -- the Achilles' heel of tech support," was originally published at Read more of Christina Tynan-Wood's Gripe Line blog at