Yogi Berra didn’t know much about contact centers, but he seems to have had great insight into the first phase of the workforce planning process when he said, “It’s tough to make predictions, especially about the future”. The simple fact is that the complex workforce planning process starts with a forecast of the future contact volume (voice inbound or outbound, email, chat, SMS, video, back office, etc.) that the contact center will encounter on a given day, at a given time, for a given type of work, and it’s “tough” to be consistently accurate in these forecasts.
The Workforce Management (WFM) system uses these volume forecasts to project the number of appropriately skilled employees that will be needed to meet optimization goals (e.g., SLA, balance of cost and revenue, a mandate to cover all contacts without limit as in 911 centers, handle tasks before a deadline, etc.) and then use that information to produce schedules, or the raw materials employees will use to build their own schedules. If the contact volume forecast is very different from the actual contact volume ultimately encountered, we are faced with one of these two very undesirable conditions: (1) unhappy customers because the average wait time is long (or deadlines are missed if back office tasks) due to insufficient number of appropriately skilled agents available, resulting in customers who abandon causing lost sales and poor CSAT or abandon one channel for another causing ‘channel escalation’ further disrupting workforce planning or (2) unhappy contact center executives because labor dollars are being spent for agents who sit idle waiting for the next contact.
Contact center workforce planners have for decades sought the nirvana of a mathematical model that can accurately predict the number of contacts that will arrive at the contact center at a specific time on a specific day for a given type of work. In fact, there are quite a few math models in use by contact centers across the globe, but two of the most popular are the Holt – Winters exponential smoothing model and the ARIMA (Auto Regressive Integrated Moving Average) model. Research shows that for some types of calling patterns, Holt – Winters is more accurate in predicting the future, and for other types of patterns, ARIMA is more accurate. Holt-Winters is particularly good at identifying seasonal patterns and projecting them into the future while being simple enough for most workforce planners to use. ARIMA is quite complex and generally requires someone on staff that has an advanced knowledge of statistics.
In either case, Yogi Berra’s wisdom is worth recounting: “It’s tough to make predictions, especially about the future”. The best statistical model can’t predict the thousands of real-world events that actually cause people to call the contact center, for example the decision by Jessica’s parents to buy her a phone for her 16th birthday, and Jessica’s subsequent call to the contact center about cell phone billing options. This undeniable truth gives rise to a useful corollary to Berra’s famous aphorism, and we could phrase it this way, ”Knowledge of the future is more powerful than statistical predictions about the future”.
Contact centers can and should use mathematical models to predict future contact volumes, and not surprisingly, the closer you are to the date being predicted, the more accurate you can be in your predictions, because tomorrow’s volume is typically related to volumes of the recent past. Models are still just statistical estimates about the future of thousands of real world events like Jessica’s that drive people to call the contact center. But, if you did have certain knowledge of the events or company initiatives that would influence customer experiences, that would give you a huge advantage in predicting the future, because you are actually making the future. For example, if you knew that on September 30 the company would start a new marketing campaign that would last 2 weeks with significant product discounts and a likely flood of calls, that’s powerful information, and you can be sure that, all other things being equal, your call volumes will go up.
Now, if your workforce planning staff knew the top 10 planned actions or events influencing call volumes, whether from within the organization or from outside, in most businesses, you have explained much of future change in call volume, and you have added significantly more accuracy to your WFM forecasting.
It is critical that the workforce management system you use allows you to apply the unique knowledge you have of your business to the forecasting process and to modify or override any input or output, so you can provide some of that “Knowledge of the future…” Your expertise with your business, industry and your relationships with others in your enterprise in charge of customer impacting events are critical additions to any method used to forecast and must be taken into account.
That’s why Aspect has designed its market-leading WFM solution with forecasting capabilities that incorporate not only the proven accurate Holt-Winters time-series predictions based on historical patterns but also the ability for the workforce planning staff to easily adjust those predictions with their own models that incorporate known actions and events in the future such as an upcoming new product release, major marketing campaign, or decline in consumer confidence.
Aspect WFM takes an educated approach to forecasting by recognizing that it’s impossible to precisely predict the future and there’s a steep cost associated with trying to achieve small additional increments in accuracy, such as having the planning staff pour over data from complex time-series models or hiring an on-staff PHD in statistics. Moreover, Aspect WFM provides workforce planners with a feature that is unique in the marketplace to help them understand the impact of not being able to precisely predict the future, both in terms of forecasted call volumes and in terms of many other factors such as future staff availability and shrinkage. This “what-if” capability allows users to easily see the impact of unexpected events on staffing and SLAs, so users know whether their planned staffing scenario is relatively impervious to unexpected events, or whether it falls apart if the forecast is off by 5%. Understanding the impact of unavoidable forecasting error allows workforce planners to hedge their bets in a world where we know, “It’s tough to make predictions, especially about the future”.
Fore more information about Aspect Workforce Management and Aspect WFM Forecasting, please contact us at 888-547-2481.
Latest posts by Eric Hagaman (see all)
- Baseball and The Myth of WFM Forecasting Accuracy - September 25, 2018
- New Survey Reveals Shifting Cloud Contact Center Preferences - September 28, 2017
- Why You Need to Upgrade Your Aspect WFM System: Workforce Management Excellence, Part III - February 6, 2017