Using Historical Data in Call Forecasting Requires Understanding History
When it comes to workforce management and scheduling, there are many factors that need to be considered. Forecasting, of course, is critical to help companies put the right resources in the right place at the right time so they are neither understaffed (which leads to long wait times and frustrated, angry customers) or over-staffed, which is wasteful.
One of the most important, as any contact center manager knows, is historical data. It tells the contact center what volume was like on a certain date at a certain time, how it was split between telephone and multimedia sources, and how many agents it took to comfortably meet this volume. But not all historical data is created equal: contact center managers must determine which historical data is relevant and which represents one-time events that aren’t likely to be repeated.
In a recent blog post, contact center services provider Blue Ocean Contact Center’s Sean Miller used some recent customer examples to illustrate events that are unlikely to be repeated again, and that therefore should not be used in forecasting.
“For folks in the online grocery business, 2013 was the year of Thanksgivukkah [when Thanksgiving occurred during Hanukkah] – an extraordinary coincidence that impacted holiday purchasing behavior in ways that won’t be repeated in 2014. (Or for the next thousand years, for that matter,)” wrote Miller.
Without the ability to recall this event, it would be easy for someone staffing a contact center next Thanksgiving to use 2013 data to overstaff a contact center by a significant amount, which would lose that business money. Not as much money, however, as Miller’s next example.
“The Super Bowl in New Jersey created higher than average bookings for ground transportation in the greater New York area in 2014, for instance,” he wrote. “Those exact conditions won’t be repeated next year when Phoenix hosts. While the Super Bowl always drives a spike in volume for our executive ground transportation clients, a Super Bowl in New Jersey produced a huge spike. It would be a mistake for us to forecast equal or greater volume for Super Bowl weekend 2015 based on what we saw in 2014.”
It’s therefore important to differentiate between historical events that are likely to be repeated – a regular holiday, a sale, a new product or service promotion or a school break that will see parents requesting time off – from an event that is unlikely to reoccur at the same time – a storm or long power outage, a plague of the flu or a major traffic jam.
Miller notes that while it’s often said that “numbers don’t lie,” in contact center forecasting, they frequently do…or, at least, they can misrepresent.
Edited by Stefania Viscusi