One of the most important components in running a contact center is ensuring you have the right agents on the phone at the right time to handle the arrival of incoming calls. There are patterns to incoming calls, but incoming calls are not scheduled and tend to run in cycles or bursts. Thus it is generally called Random Call Arrival. Having an accurate forecast allows you to estimate the call patterns and ensures proper coverage to meet your service level objectives. Once you have built your forecast, you then use Erlang or other more modern methods to build the schedule.
There are five primary ways to comprise the data required for an accurate forecast:
Rolling Forecast – A Rolling Forecast uses the averages from the last weeks and averages them to predict the call volume for the upcoming week. Each time the forecast is run, it is for the most recent past several weeks. This type of forecasting is best for companies that have daily fluctuations, with call patterns that match other weeks. For a quickly growing company, most automated workforce management systems will allow the user to add a percentage increase for company growth. This is the most commonly used forecasting method and typically only uses the 6 most recent weeks of history.Static Forecast – A Static Forecast averages or uses specific weeks in history to best model the upcoming week. For example, a week that has a Monday holiday causes calls to be heavier on Tuesday than a normal week. So rather than averaging the last several weeks together, it is important to average the last several Monday holiday weeks to gain a better estimate of the call volume. As these dates may be more spread apart, taking into account any growth of the company is important. You should use at least 4 past weeks to help make an accurate forecast.
Date Range Forecast – A Date Range forecast uses specific days of the year to best plan for upcoming similar dates. Companies with monthly billing cycles may have higher call volumes in the first 5 days of the month. The first 5 days of a month are not going to be the same weekday month to month, however calls spike on the specific date of the month and not day of the week. This can also be used to forecast holiday weeks like Christmas and New Years that do not fall on the same days of the week, year over year. When using specific days of the year, you may need to use multiple years of data to get an accurate forecast. So making sure you keep contact center data long term is important for this type of forecasting.
Excel Data Forecast – Many contact centers keep their call volume history in Excel. They may have been able to export the data into Excel from their phone system. It is best to have 15 minute intervals of data (number of calls presented and average handle time) for each skillset. It is common for outsourced contact centers to obtain this type of information from prospective clients. Uploading the spreadsheet of data into the workforce management solution allows the company to cost the addition of the business and to see how it may impact other lines of business to make hiring decisions. This method allows for mass import of relevant data that does not exist in an integrated solution. In nearly all cases, uploading Excel data into an automated workforce management solution is used to plan for adding agents or taking on additional business.
Manual Data Entry – If nothing else is available, manually creating the data may be required. This method is best used to create new agent requirements or “what if” type calculations. For example, you want to add a skillset and can assume a certain call volume will transition to that skillset. You would need to manually create the data for the new skillset. This can also provide an option to create scenarios for taking on additional workload or transitioning coverage to optimize agent resources. Manually entering call data for the week will need to be based on 15 or 30 minute intervals. Due to the level or work associated with this method, it is common to enter only one set of data that will be used as the forecast instead of multiple sets of data that will be averaged.
A good forecast starts with sourcing the right data. Understanding what data you will use and how it will affect your agent requirements and service level is part of the art of workforce management. You want your forecast to best represent the type of call patterns you expect in the upcoming week. Through these five different methods, it is possible to pull together the best data to represent the upcoming requirements. For more information on Forecasts and Workforce Management Solutions, go to www.cxmrecord.com.