Determining Call Center Staffing Needs

Three foundational building blocks need to be optimized to consistently deliver engaging member experiences over the phone: people, processes and leadership.

Our journey of nine articles will cover each in detail. Part one discussed hiring top performers and performance management. Part two covers processes. This is the sixth article overall and the third on processes. Specifically, in this article we’ll tackle how to calculate the number of agents needed to staff your credit union’s call center.

After a few re-writes, we decided against publishing a dry, basic overview of this topic that would ultimately become a really good sleeping aid (we apologize to those already dozing off!). As a result, and for your entertainment pleasure, we’ve written this article in dialogue format between a senior leader and his direct report (call center grand poobah). It’s a dialogue you’ve likely encountered or will at some point in your career.

Grand Poobah: “Boss, we need to hire additional agents to properly staff our call center for 2017.”

Executive [shaking his head, thinking not this again… but trying to be open minded]: “What brought you to this conclusion?”

Grand Poobah: “We aren’t reaching our target service levels.”

Executive [in a stern voice]: “I admit I don’t know much about our target service levels, but we have 10 agents and each agent takes about 50 calls a day. If each call is six minutes in length, that’s five hours of work per agent. What the heck are agents doing for the other three hours?”

Grand Poobah: “It isn’t that simple boss. For starters, calls come in randomly – not sequentially. Therefore, you can’t calculate staff that easily.”

Executive: “Well then, how do you calculate staff?”

Grand Poobah: “I’ll keep this as simple as possible. Let’s assume we receive, on average, 10 calls between 9 and 10 a.m. and that each call is six minutes (60 minutes total). You with me?”

Executive: “Yeah.”

Grand Poobah: “Assuming no breaks…”

Executive [interrupting]: “You know when I was younger, we never really took any breaks. Work ethic was much different.”

Grand Poobah [shaking his head]: “Based on your approach, you’d assume that one call center agent should handle all these calls between 9 and 10 a.m., right?

Executive: “Yes!”

Grand Poobah: “OK. Now consider the impact of your decision to have just one agent on the phones when six of these 10 calls arrive between 9 and 9:20 a.m. Odds are at least half those callers are going to have to wait awhile to reach a live agent.”

Executive [being sarcastic]: “Why don’t we tell members to call at certain times so we can save them money in the long run?”

Grand Poobah: “Let’s look at this from the agent’s perspective. For starters, being that the agent sees calls in the queue, he/she will then attempt to get through calls as quickly as possible so members don’t have to wait. Consequently, not everyone will receive the “WOW” service we are wanting to deliver.”

Executive: “That’s not good.”

Grand Poobah: “There’s more ... Some of those members, at least those who don’t abandon after waiting a while, won’t be in the brightest of spirits when they finally reach a live agent.”

Executive [being somewhat sarcastic]: “We should play relaxing music while on hold.”

Grand Poobah: “Good idea, but there’s more. Let’s assume all callers patiently waited and ultimately reached a live representative and, after their call, received a survey.  What are the odds your agent will score really well on all surveys?”

Executive: “Probably will have some score below average, which isn’t what we want.”

Grand Poobah: “Exactly. And what if, during that same time period, one of the various systems our agent had to log into was down and, consequently couldn’t resolve a member issue? Not all members are forgiving you know…

Executive: “Dang systems!”

Grand Poobah: “The potential long-term impact, should this trend continue, is that our agents’ morale suffers due to issues they couldn’t control. Then they start coming to work late, take longer breaks, etc. Along the way, we’ll also upset some members – even lose a few. Now multiply this by a much larger volume of calls and you see how critical it is to allocate enough resources at the right time – especially important since people are our biggest expense.”

Executive: “OK. I get it. But how do we ensure we aren’t over hiring? Other executives already view this department as a cost center.”

Grand Poobah: “Well they are wrong. This department is building our credit union’s brand. Why do you keep throwing money at the branches when we interact with more members than any other branch? Anyhow, before I get into staffing models, I wanted to let you know that I’ve forecasted call volume based on a three-year history of calls and took into consideration upcoming marketing campaigns, system maintenance and our online banking conversion. I have forecasts down to 30-minute increments. So I’m not just pulling numbers out of thin air.”

Executive: “How accurate is this forecast? Hopefully more accurate than our local weather man!”

Grand Poobah: “Well, it’s impossible to be 100 percent accurate, but we shouldn’t be off by more than 5 percent. Let’s get into Erlang C.”

Executive (with a smirk): “Is he the brother of the guy who developed Hi-C?”

Grand Poobah: “Don’t quit your day job, boss.”

Grand Poobah: “Erlang C is a model developed in 1971, by A.K. Erlang. He was a Danish mathematician and engineer who invented the fields of traffic engineering and queuing theory – more specifically a solution to the problem of how many phone operators were needed to handle a given volume of calls. Most call centers use this – or variations of the model – to determine staff and scheduling.”

Executive (with a smirk): “Kind of like using an interest rate risk model.”

Grand Poobah: “Somewhat. While it’s not a perfect model, because it tends to slightly over-estimate staff (one reason is that it does not assume any abandoned calls), it’s still a widely used model. In fact, even many workforce management/planning solutions use the model with variations to make it more real-world applicable. Worst-case scenario is that we are at least basing our assumptions on a model rather than just randomly guessing.”

Executive:  “So how does it work?”

Grand Poobah: “Well, one approach is to use an online model and plug in key inputs and the model will output the information we’re looking for. To do this, there are a few things we need to know. “

Executive: “Such as?”

Grand Poobah: “First, our service levels. A service level is a measure used to refer to our team’s responsiveness to members. Specifically, the percentage of calls we strive to answer in a matter of seconds. While no industry standards exist, it’s most common for call center service levels to be 80/20; meaning the goal is to answer 80 percent of the calls within 20 seconds. Our service level is 85/15.”

Executive:  “How did we determine that?”

Grand Poobah: “There are many approaches (not one right answer) to do this. The approach we took was surveying our members about their expectations. Other call centers sometimes determine service levels by benchmarking financial institutions similar to them. Others may use a quick rule of thumb based on the tolerance they have for abandon call rates (everybody defines these differently, too). Typically, the higher your service level the lower the abandoned rate. For example, if you consistently reach 85/15 or in that range, your abandoned call rate will typically be in the 1-2 percent range. 80/30 ranges typically see an abandonment rate of 3-4 percent and so on. “

Executive:  “This is getting complicated. What else do we need to know?”

Grand Poobah: “Average talk time is another input. This is simply the average duration of each phone call our agents make. And this is where effective training and efficient systems are so important. The quicker agents can diagnose the problem and navigate through (often several) systems, the more time they have to build rapport with members.”

Executive:  “Is there more?”

Grand Poobah: “Yes. Another input is average handle time.”

Executive:  “Let me guess. This is the amount of time for the entire call.” 

Grand Poobah: “Close. This is the overall amount of time that is necessary to answer a call and includes everything from inputting notes into our system to completing paperwork (after call work) – basically whatever is needed to complete the transaction.”

Executive:  “Doesn’t this pull people off the phones?”

Grand Poobah: “Yes. It’s impossible to eliminate. However, we can do a lot more to minimize after-call work. Streamlining processes will help ensure we aren’t overstaffing.”

Executive:  “I’m all ears!”

Grand Poobah: “This is where journey mapping and process improvement come into play. A good exercise is to “staple yourself” to a common transaction and walk through the entire process. Then figure out how to streamline the process and create a better overall member experience. Basically we need to do this for the most common calls that come in through the call center (there are a bunch of them). Mapping all this out will also expedite the learning curve for new hires.”

Executive:  “Now can we plug into the model?”

Grand Poobah: “Yes. However, what the model outputs is just the start. There’s also the issue of shrinkage.”

Executive:  “Shrinkage?”

Grand Poobah: “Shrinkage (or rostered staff factor) is a measure of how much time your agents are spending off the phones due to paid lunches and breaks, time off, meetings, training, etc.”

Executive:  “We allow too many breaks. When I was your age, we never took any.”

Grand Poobah: “As I was saying … calculating shrinkage provides call centers with the minimum staff needed to schedule over and above the base staff to achieve desired service levels. Here’s how you would typically measure shrinkage”:

(Grand Poobah pulls out whiteboard and starts writing the following)

Assume 28% shrinkage: Scheduled Staff = 100 ÷ 0.72 = 139 staff needed

Grand Poobah: “In sum, Erlang C and shrinkage can help us determine staff.”

Good luck with your next call center staffing conversation and stay tuned for next month’s article about how to create a learning environment.

Prior to starting his new company, Bill Stavros spent over 13 years of his professional career in the credit union industry, most recently as VP/marketing & member experience at $513 million Proponent Federal Credit Union, Nutley, N.J. At Proponent FCU, Stavros was responsible for conceptualizing, architecting, developing and deploying a new vision for delivering customer service for the inbound call center, based on his learnings from highly regarded service oriented companies such as Zappos, Ritz-Carlton, Disney & REI. Now with Blueprint Interactions, his goal is to help other credit unions achieve similar results. He can be reached at bill@bpinteractions, by calling 888.757.8338 or at www.bpinteractions.com.