By the Numbers: Contact Center Metrics

By Linda Formichelli

If train A leaves Boston at 3:13 p.m. going 75 miles per hour, and train B leaves New York City at 2:24 p.m. going 71 miles per hour, what time will they meet? You probably thought that once you’d graduated from high school you were through with having to solve this sort of problem. But now that you’re in charge of a contact center, you may be struggling with math once again. Should you measure the percentage of calls answered in a certain amount of time, the average speed of answer, the abandonment rate, or all of the above? And once you’ve finally decided, how do you implement those metrics — and how will they affect other aspects of your business, such as staffing? It’s high school algebra all over again.

PICKING AND CHOOSING

Deciding which metrics to measure is “the start of some of the greatest errors in the trade,” says John Cockerill, vice president of the Taylor Reach Group, a contact center consultancy in Toronto, Ontario.

Most contact centers measure service level, usually defined as the percentage of calls answered in a certain amount of time. A service level of 80/20 — a common goal for multichannel merchants — means that 80% of calls were answered in 20 seconds or less. While average speed of answer (ASA) and abandonment rate are also important metrics, the service level statistic can give you a good idea of your performance in those related areas.

“Our service level is probably the most comprehensive indicator of all the other ones, so that’s the key one we focus on,” says Tena Perrelli, senior customer contact center manager at Burlington, VT-based Gardener’s Supply Co. and Dutch Gardens. “When our abandonment rate goes up, service level goes down. The same with call length — if it’s not according to what you budgeted it to be, it will negatively affect your service level.”

ASA is widely used, but some experts believe that contact center managers rely too much on this metric. “It’s misleading,” says Brad Cleveland, president of the Annapolis, MD-based International Customer Management Institute (ICMI). “It’s mathematically correct, but it’s skewed.” For example, an impressive-sounding ASA of 15 seconds can conceal a longest wait of an unacceptable three minutes. “For that reason,” Cleveland says, “we don’t recommend average speed of answer [as a metric]. It’s an alternative if you don’t have service level, but service level is better.”

Abandonment rate is another metric that’s widely used — but it can be misleading as well. “There’s an irony with abandonment,” says Cleveland. “If you have a good service level, you can see abandonment going up in shorter increments of time.” Say you have a consistently good service level, and a customer is waiting on the phone when someone walks into his office. Chances are, he’ll bail out on the call, knowing that he can always come back and expect a quick answer. On the other hand, if you’re known to have a poor service level, that customer won’t hang up no matter what. The result: a low abandonment rate that, ironically, reflects a poor service level.

Another metric to consider is blockage: the number of calls that can’t get into your switch because you don’t have enough capacity. In fact, says Cockerill, abandonment rate and blockage rate are key components of service level, which gives you a stat like 80/20/3/0 — 80% of calls answered in 20 seconds or less, with 3% abandonment and zero blockage.

In the multichannel environment, you’ll also have metrics for two types of nonphone channels: those that require a response as they come in (such as text chat and instant messaging) and those that don’t (such as e-mail and postal mail). To gauge performance within the channels that require immediate response, you can use metrics similar to those you use for phone calls. The others require different response metrics and goals, such as 100% of e-mails answered within 24 hours, or 70% of mail answered within two days and 100% answered within four days.

SETTING STANDARDS

So let’s say the service-level metric is the most important (though not everyone agrees — see “Counterpoint: why service level isn’t an ideal metric,” on page 45), with ASA, abandonment rate, and blockage rate playing supporting roles. The next question: How do you determine what service level to aim for?

It depends on your value proposition to customers. If you differentiate your company based on service, you’ll need to back that up with a higher service level, says Cleveland.

In his book Call Center Management on Fast-Forward: Succeeding in Today’s Dynamic Customer Contact Environment, Cleveland notes that for catalogers that want to be on the high end of the service-level scale, 90/20 (90% of calls answered within 20 seconds), 80/15, and 90/15 are common service-level goals, with abandonment-rate goals of 1%-2%. Service levels of 80/20, 80/30, and 90/60 are considered average and would likely yield an abandonment rate of 3%-4%. A modest service level might be 70/60 or 80/120. These humbler measures are more common in technical support centers and government organizations, where the abandonment rate can be as high as 10%-15%.

STRIKING A BALANCE

You say your business is all about service; it’s what you promise your customers, and it’s what you pride yourself on. So should you overstaff your contact center in order to avoid a single missed or abandoned call — even if doing so will require amazing feats of staffing prowess? In other words, how do you balance your contact center metrics with staffing and cost?

Most consultants and practitioners agree that it’s better to aim for a higher service level and be slightly overstaffed than to be understaffed with a lower service level. “We had a consultant here a week ago, and he showed that you can err on the side of being overstaffed, and in most instances the sales you save from not losing calls offsets the sales you lose if you’re understaffed. It’s pretty dramatic. So we’re trying to apply that when we’re making cost justifications for increased staffing,” says Gardener’s Supply Co.’s Perrelli.

Opting for a lower service level in order to avoid overstaffing can cost you money in another way as well, says Cockerill. If you have a long hold time or a cumbersome interactive voice response (IVR) menu at the front end that takes a long time to navigate through, once the caller finally reaches an agent he will spend time — generally 15-45 seconds, Cockerill says — complaining about how terrible the service is, turning what would have been a short, efficient call into an unacceptably long one. This, in turn, increases the degree of bad service the other callers in the queue will get — which translates into increased costs.

“Most people presume that good service costs more on a sustained basis…but in fact, our evidence shows that it’s quite the opposite,” says Cockerill. “Bad service costs you a lot more.”

Of course, there is a limit to how much you’ll want to pump up your service level by adding agents. “If your service level is 100%, you’re going to be overstaffed,” says Perrelli. “People hang up for lots of reasons, and they will call back depending on their reason. What you want is to not lose calls because people are tired of waiting.”

All these numbers and calculations come down to one thing: keeping customers happy so that they’ll buy (and continue to buy) from you. Correct staffing and proper service levels will save you money on overstaffing; at the same time, keeping customers happy will lead to more profits.

Unfortunately, however, the customer often gets lost in the mix of metric numbers. “There’s generally a 30% difference between the internal scoring measures vs. the external scoring measures of how well your customers think you’re doing,” says Cockerill. “And 99% of the time, they think you’re doing worse than your internal scores. That’s because we measure what we think is important, not what the customer thinks is important.”