Intraday Automation: The Next Sea Change

In an earlier blog, I talked about the challenges faced by call centers in the mid-1970s and how technology has changed the game.

My very first call center back in 1973 started with a PBX (Private Branch Exchange). Customers called in and were passed around until an agent became available. Or, if all incoming lines are busy, customers would get a busy signal and have to call back.

Scheduling was fairly straightforward and done manually on a piece of paper.

Automatic Call Distributors brought some science to how calls were distributed. For example, which number customers used to call in would determine their queue and ultimately, which agents would take their calls.

Then Interactive Voice Response units (IVRs) came along, allowing customers to enter personal information like their account or social security number, which would put them in the queue for an agent with the specific skill set required to help them.

This technology worked well for centers with one number, one ACD, one queue and one skill set. But imagine a major airline today, for example, with multiple contact centers and multiple numbers for customers to call in depending on status (a separate service number for silver, gold, platinum and diamond customers, for example).

And each of these four queues has a different service level. For example, “silver” customers might wait longer than “diamond” customers. So now you have multiple queues that require multiple skill sets in multiple centers.

Situations like these quickly become so complicated that only a handful of workforce management systems can effectively schedule them even today.

Staffing Variables

When it comes to forecasting how many agents are needed in the contact center at any given time, there are also many variables. How many answered calls are there? How many abandoned calls? How many customers have received a busy signal? How long does each call last?

One way to measure this is actual “talk time,” which the ACD can provide, but there is also something called “Customer Resolution Time,” which is talk time plus after call work.

Assume you have 200 calls in an hour. Each call takes five minutes and the service level you want is “90% of calls answered in 20 seconds or less.” Using a conventional Erlang C table – a standard in the contact center industry – you would forecast 23 agents for this situation. However, with Erlang C, call arrival rates are random. You might have more calls in the first two minutes than in the last two minutes, for example.

Also, Erlang assumes an infinite queue in which all calls are eventually answered. As you know, this is unrealistic.

The truth is, as long as you have just one queue, one skill set, and one center, Erlang C works fine, and is used by lots of companies today. But when you have multiple sites, skill sets and queues, Erlang C is no longer adequate.

Typically, for sophisticated complex situations, you almost need some sort or simulator to determine the arrival rates and number of agents you need.

In my earlier example, I used Erlang C to forecast that you would need 23 agents to handle 200 five-minute calls an hour with a service level of 90% of calls answered in 20 seconds or less. But if you were to add one minute of after-call work to every call, you would need 27 agents to handle this volume.

And there lies the rub! If you just manage talk time and don’t take into account after-call work, you won’t have enough agents on the floor.

Another variable – perhaps the most important variable in forecasting – is percent of occupancy, not to be confused with utilization. What are agents doing when they are logged in? Are they talking to customers? Completing after-call work? How long are agents sitting idle between calls?

Utilization is not the same as occupancy because utilization must be divided by paid time. What are your agents being paid to do? To be logged in? To take two breaks a day? To complete a certain number of training and coaching sessions?

If your occupancy rate is 90% when agents are logged in, they are idle 10% of the time, but aren’t getting any breaks or completing any training or coaching.

Measurements Drive Behavior

A few years ago, I was in a large contact center that measured all kinds of metrics. When the queue reached 50 customers, the superintendent stood up in the middle of a room of 400 agents and yelled as loud as she could, “Quick! Contact!”

Immediately, all calls dropped out of the queue.

I asked her what just happened and she said, “We are measured on service level. Eighty percent of calls have to be answered in 20 seconds. Anytime I see that our service level starts to go down and we have 50 calls in queue, I am supposed to let everyone know so agents can tell the customer on the line that their computer just went down and they should call back later.”

When I told the company’s vice president about what I had seen, he didn’t believe me. I told him, not only is it happening in your center, you encourage it! If you measure agents on service level and bonus them on service level, they will do anything to make the number. If taking care of customers is really the most important thing, measure First Call Resolution and forget everything else.

Intraday Automation: The Next Sea Change

For the CEO, the main concern is usually are we making any money?

If 65-70% of your revenue is direct expense, are you paying agents to sit idle, waiting for the next call? I have seen some companies check direct expense to revenue every half hour and if the numbers aren’t right, they send agents home (even though they will need those agents back two hours later when volume spikes).

At the same time, most CEOs will say that customer experience is the most important metric.

In today’s contact center environment, multiple centers, queues, channels, skill sets and technologies – combined with huge amounts of data – have made managing all of this effectively extremely difficult. It’s complicated – and it’s all happening in real time.

As in life, unpredictable things happen in the contact center every day and decisions have to be made quickly, based on multiple variables. With all of our sophisticated call routing, IVRs, WFM systems, reporting, quality and learning management systems, new technologies like intraday automation sits in the middle and turns all of this data into something actionable.

These technologies fill the gap, automatically making the decisions that are best for your agents, your business, and most importantly, your customers.

About the author

John Englund

John is a copywriter at Intradiem. He has a background in print and broadcast journalism and digital marketing with emphasis on technology.

Similar Articles