Improved Call Volume Forecasting Can Lessen Scheduling Problems for Your Contact Center
How useful a workforce management (WFM) software-created schedule is will greatly depend upon the accuracy of the call volume forecasting predictions it makes for your contact center. If incorrect assumptions are made about the anticipated call volume, the software will fail to accurately calculate the staffing required to handle that workload. The impact on the bottom line can spell disaster.
When the call volume forecast is too high, overstaffing occurs and too many agents are scheduled. This results in wasted labor expense, boredom and lower agent engagement which in turn negatively impacts customer experience and satisfaction levels. When the workload forecast is too low, understaffing occurs causing longer average speed of answer, more abandoned calls, lost sales and lower revenue. Either way, staffing mistakes from poor call volume forecasting eat into your profits, alienate and drive away customers and cause long-term harm to your business.
Here’s the issue: all forecasting tools are not created equal. Only the most sophisticated WFM solutions can perform correlated forecasting; i.e., forecasting that can account for special events like catalog drops, advertising campaigns, sales and other promotions, holidays, etc. Such events can cause dramatic fluctuations in call volumes, as well as in e-mails, texts, faxes and online chat sessions in multi-channel contact centers. Because of this, the forecasting functions of any WFM software suite should be a primary consideration in your purchase decision. Here are several factors to keep in mind, some steps to maximize call volume forecasting accuracy, and the significant contributions the right forecaster can have on contact center profitability.
The Value of Pattern Recognition
There are two basic methods for call volume forecasting in a contact center environment: Exponential Weighted Moving Average and Historical Trend Analysis. Both methods take into account historical data collected from the center’s ACD, and both account for growth trends in their calculations.
The Exponential Weighted Moving Average computes the average call volume over a set period of time, and bases its projections on a formula that assigns more weight to more recent activity. This technique is more effective for centers where there is only slight change in call volume and patterns, places like help desks and technical support organizations, but becomes less effective as trends change. It cannot predict a continuation of trends during periods where volume generally increases/decreases, and is unable to associate volume changes or call arrival patterns with certain events (pattern recognition).
In contract, Historical Trend Analysis can accurately predict the continuation of trends, and with more advanced algorithms can also include pattern recognition allowing you to fine-tune workload forecasts for special events like advertising campaigns, sales promotions and national or state holidays. With each occurrence of a particular event (or similar event), the call volume forecasting is adjusted automatically to reflect the increase or decrease in incoming calls or other work caused by similar past occurrences, like a historical drip in volume on the 4th of July of 40 percent.
In contact centers where workloads are prone to significant fluctuations due to marketing activities and other definable variables, Historical Trend Analysis is the best way to guarantee proper staffing levels because it is the only method that is able to include complicated historical trends in its calculations. Without pattern matching abilities for predicting customer behavior for different events, over- or understaffing risks increase dramatically.
Data Validation & Mapping Historical Data To Special Events
When using a WFM solution that employs pattern recognition, a critical step that cannot be overlooked is regular, ongoing data validation. Analysts must review the center’s ACD data on a weekly basis at a minimum (or preferably on a daily basis) to determine if there are identifiable causes for any and all spikes and drips in call volume.
Typically, unusual patterns in the ACD data will be related to identifiable events such as catalog drops, advertising/ direct mail campaigns, TV advertorials, sale or discount offers, competitor promotions, pay periods, billing cycles or holidays. Some fluctuations may be linked to other external factors like the Super Bowl, Olympics, or severe weather events like snow storms, hurricanes, extreme heat waves, etc.
If a given spike or drop in call volume was triggered by a recurring special event, analysts can instruct the forecaster to interpret the data set accordingly. If a fluctuation is the result of a one-time event (such as a product mention on Oprah or The Today Show), analysts can tell the system to ignore that data set when forecasting. These instructions are key to achieving high levels of call volume forecasting accuracy.
Defining Special Event Attributes
To further increase forecast accuracy, some call volume forecasting tools also make it possible to describe every event that causes a fluctuation in detail by using attributes. One catalog drop might contain 10,000 pieces sent to women between 20-35 in Southern California, while another targets 5,000 older women living in the Midwest. By defining these characteristics in the system, analysts make sure that the different call patterns created by each different catalog drop will be ‘remembered’ and factored in call volume forecasting in the future when similar mailings go out.
WFM systems with the most sophisticated forecasters can search for historic trends paralleling those expected from upcoming events. This can be done by exact match (a specific guest host on a TV shopping channel) or by a range of values (products priced between $50 and $100 being featured on a TV shopping channel). This assists in the correlation of past and future special events. There will be a significant difference in response to a $200 piece of jewelry being featured on a TV shopping channel versus a $2,000 piece of jewelry. Only a call volume forecasting tool that allows this type of information to be input and used in its calculations can factor the difference when generating workload forecasts and schedules.
The Impact of Improved Call Volume Forecasting on Call Center Revenue
Once historical patterns can be identified, attributes get assigned to events and upcoming events are logged into the system, call volume forecasting can be completed with a much more accuracy than with tools that lack these abilities. Requirements for staffing, as a result, can be predicted much more precisely. The significance of accurate forecasting can be identified by understanding the consequences of an inaccurate one.
Say a workforce management package underestimated a center’s call volume, therefore also underestimating the center’s required staffing levels to the point where 100 out of 1,000 callers hang up before an agent speaks to them. In a sales environment where an average order is merely $50, that equates to $5,000 in revenues lost per day, $150,000 lost per month, and a mind-blowing $1.8 million in revenue lost in per year. Best case scenario, the lost sales cut into the center’s profits; worst case scenario, they have the potential to ruin the entire business.
There are many other elements that factor in the equation dictating how effective a given WFM package is/will be. These include the software’s sensitivity to agent skill sets and work rules, its real-time adherence monitoring functionality and its ability to compute requirements for staffing based on highly detailed user-defined services levels, varying from mean time to answer, to percentage of busy signals and abandoned calls that will be allowed.
But all that is a moot point if the software’s forecasting tool doesn’t match the center’s needs. Since all agent assignments are based on estimated call volumes, a WFM solution that has inadequate forecasting ability is comparable to a weatherman with out-dated technology. Both parties will end up with a disproportionate number of incorrect predictions. For call centers that rely on accurate staffing in order to accomplish their work, selecting the best call volume forecasting solution can make all the difference…and ward off an almanac’s worth of stormy days.
Pipkins, Inc. (PIPKINS), founded in 1983, is the leading supplier of workforce management software and services to the call center industry. Its Vantage Point product enables managers to solve the complicated operational issues in today’s multi-faceted call center environment. It offers a rich and robust feature set, as well as an open design that allows for the complete integration of emerging CRM technology. PIPKINS’ systems forecast and schedule more than 100,000 agents in over 300 locations across all industries worldwide. The company is headquartered in St. Louis, Missouri.