Call center forecasting is tough as nails to do well with a spreadsheet, unless you know how to program a service level prediction formula like this into Excel:

z_dk =  ρ_{qd ,k} +φ  d + ε_dk;         ε_dk ∼ N(0, σ²),

Not in your wheelhouse? It’s not in mine either. Fortunately, you don’t really have to do that, if you’re using a Workforce Management System that does it for you.

When it comes to call center forecasting, call volume is what many people think of. We look at inbound call loads and try to staff accordingly. But to measure volume on its own is to call center forecasting what an aching knee is to weather forecasting. Accurate schedule forecasts don’t come from the ringing phone, but from the historical analysis of customer behavior.

I thought I’d share a little insight into some of the data advanced WFM tools use to develop staffing forecasts.

When do your customers call?
Good forecasting tools analyze activity on an hourly basis (sometimes even more frequently). It’s not enough to know that your busiest periods are the afternoons on Tuesday through Thursday. “Afternoon” is a very long—and potentially expensive—timeframe. A more complete and detailed understanding of hourly patterns opens the door for more efficient scheduling.

Why do your customers call?
Just as volume varies from hour to hour, so does caller intent. Forecasting analyzes mission patterns. That includes routine and specialized tasks: when people check their balance; when they pay their bills; when they call for returns; when technical support is high; when the most upsell opportunities occur . . . and others.

Who do your customers need to talk to?
Analyzing volume and mission allows you to forecast not just staffing levels but skill levels as well. If you find customers are most receptive to offers in the evening, you bring in agents with high upsell success. (This is also a key indicator of how well your team is cross-trained, something that forecasting will help with as well.)

What influences your customers’ behavior?
Some external events are unpredictable, but most aren’t. Billing cycles, promotions, media attention product recalls, seasonal changes in clock time, and many other factors influence customer behavior, and that influences staffing levels.

How can technology handle your customers’ needs?
Your staffing forecast has to include how much staff you won’t need by applying these same criteria to your IVR and other supporting technologies. Your IVR should be able to successfully take on some of the load. (Check out our IVR Blog for a deeper dive.)

I hope I’ve made the case that call center forecasting is based on a rich, complex and very multidimensional analysis. But it’s simple to do with Workforce Management systems that do all the hard calculus for you. Many modern WFM vendors offer a forecasting module. It’s a standard feature with our solution (you can read more about that here).