In 1917 Danish mathematician A.K. Erlang developed a model for predicting how many circuits the telephone company would needed to serve the city of Copenhagen. The model became known as Erlang C and eventually became the mathematical foundation for both automated and manual forecasting models used to estimate labor requirements in call centers. While useful, the approach had some significant limitations like the assumption that calls will never be abandoned or will never be blocked by busy signals. But perhaps the biggest limitation is that the math was designed for a world when telephone was the only practical way to communicate over a long distance.
So how does this to work in a multichannel world? Absent an ACD, how would the algorithm even know how many chats, texts, and other non-voice communications are waiting in queue to be serviced? And then there’s email. Erlang C makes the nettlesome assumption that communications which start in an interval must end in the same interval. While some email responses can be handled almost immediately, in most cases consumers choose email because they expect the agent will do some research before responding. There are some software tools for scheduling chat and email specialists, and that may be a solution if the contact center doesn’t mind separate and siloed forecasting systems for each of its three major channels.
To further complicate matters let’s assume that the consumer uses multiple channels during the same session. This is not unrealistic. The consumer may initiate the session by selecting the chat option from the corporate website, and then upon finding that option unsatisfactory, insist on being connected to a live agent. Or not. In a growing number of contact centers, multiskilled agents can seamlessly communicate in different channels and even different languages. In the words of the famous songstress, Avril Lavigne, “It’s complicated.”
Today leading workforce management vendors deal with this complexity by employing PhD level mathematicians to construct dozens of forecasting models, each based on different inputs and assumptions. This is all good, but it only adds to an already thorny situation as each algorithm has to be monitored and tweaked. And any deficiencies in individual models can impact other models. Aspect Software thinks they have a better idea. In March of this year, Aspect was awarded a patent for a Workforce Management Multichannel Scheduling System. This includes a master forecasting engine that can model and simulate the interactions of digital communications including chat, email, instant messaging, social media and other text-based communications to create best fit schedules and forecasts to meet SLAs. With Aspect’s Multichannel Scheduling System, maybe multichannel forecasting won’t be so complicated after all.
Latest posts by Dick Bucci (see all)
- Millennials and the Work Ethic - November 1, 2018
- “It’s Complicated” (But Aspect has an answer to multichannel forecasting) - July 16, 2018
- Are You Ready for the Gig Economy - April 24, 2018