#contactcenterworld

It does not matter whether you are using spreadsheets, googlesheets, or an automated workforce management system the concept to forecast a campaign (Weekly Schedule) will remain the same as per the best standard practices of the contact center.

Content Credit of the article goes to the book written by Penny Reynolds on Call Center Staffing - The Complete, Practical Guide to Workforce Management.

The purpose of the Forecast:

The purpose of the forecast is to predict workload in terms of the number of calls expected and the time it will take to handle them. The forecasting process involves a statistical interpretation of historical data to predict future workload. There are many different situations in the call center environment that require a forecast to be done.

The most common scenario for which a center will forecast is simply normal, day-to-day operations. But a forecast may also be required for special situations such as:

Opening a new center:

A forecast may need to be generated for an expansion or opening of a new center. In some cases, data are available from other call center operations to predict call workload for the new site. Other centers may have no information on which to base a forecast, requiring the center to simply benchmark against operations to begin the forecasting process.

Merger or acquisition:

As a new site is acquired, or multiple sites are merged together, it may be necessary to reexamine the forecasting process to determine how calls will arrive at a consolidated single site or be shared at multiple sites.

Change in Operating Hours:

As call centers expand hours of operations, a new forecast will not only be needed for the additional hours of availability but also for the previous hours of operations as callers shift their calling behaviors to match the new hours. Cutting back on hours will have a similar but reverse effect.

Implementation of the new technology:

Any technology that affects the flow of calls into the center must be considered. For example, the implementation of interactive voice response (IVR) technology may be offered a significant portion of calls that can be handled in a self-service mode. Call volume may be reduced and handle times may be affected as agents are left with the more complicated types of calls to handle over the phone.

Whatever the reason, it's important to understand the basic principles behind workload forecasting and how to apply them to accurately plan call center resources.

Types of forecasting approaches:

Once the data has been analyzed and adjustments made to the historical information, then it's time for the next step. The next stage in the process translates the raw data into a prediction of what's coming for a future month. There are many different types of forecasting approaches used by call centers today to predict call workloads. The most commonly used approaches are:

Point Estimation:

The simplest approach to predicting what an element of data will be in the future is to simply use an equivalent point in time in the past and copy it. This approach can be illustrated by thinking about another type of forecasting - weather forecasting. With a point estimation approach, a prediction of the high temperature for August 1st of the coming year could be derived by simply using a number from last August 1st as the estimate.

This point estimation approach has obvious limitations as a forecasting methodology. Simply selecting one point in time to represent another future period is problematic in that the selected data may have not been representative of the time period to be forecast. In the weather example, the first part of last August may have been unreasonably warm or cool, and not at all vindictive of what early August temperatures are the majority of the time.

When applied as a call center forecasting model, the point estimation approach is weak for many reasons. There is no guarantee that the past information being used as the predictor is valid in the first place. But even if deemed valid for the past, the data does not reflect any long-term change that may be an upward or downward trend in the actual call history. Therefore, the point estimation approach is rarely used in call center forecasting.

Averaging Approaches:

A step up for point estimation is the averaging approach, where several points of data are used as a predictor for the future. This approach is clearly better since multiple points of information are used, reducing the possibility that one invalid piece of information could drive the forecast in the wrong direction. One can see how this approach would serve as a better predictor by thinking back to the weather example. Rather than just talking about last year's high-temperature information, meteorologists typically take the last 50 years worth of information to determine an average high and low temperature for a particular date.

The most common Average Approaches are Simple Average, Moving Average, and Weighted Average.

Regression Analysis:

Another type of forecasting approach is useful where future call volumes are dependent upon an event or variable in addition to the normal historical influences. For example, suppose a catalog company has kept careful records of the number of catalogs mailed and the corresponding number of incoming telephone calls. This information will be a primary driver in predicting the number of calls to be expected in the next several months based on the planned number of call logs to be shipped.

Regression analysis is a tool for making this type of prediction where it is believed the incoming workload for the call center depends upon another specific variable. It is used where one quantity (Call volume in the above example) has a dependent relationship to another variable (planned catalog shipments). Call volume in this example is called the dependent variable while the number of catalog shipments is the independent variable. Regression Analysis can be used to develop a numerical relationship - a formulation in other words - between the dependent variable and the independent variable.

Time Series Analysis:

The most accurate approach for call center forecasting involves a process called time-series analysis. This approach takes historical information and allows the isolation of the effects of trends (the area of the change) as well as seasonal or monthly differences. It is the approach used in call centers and serves as the basis for most automated workforce management forecasting models. The basic assumption is that call volume is influenced by a variety of factors over time and that each of the factors can be isolated and used to predict the future.

Time Series Analysis step further to isolate the effect of trend in the data, rate of change, upward and downward trends, etc.

Time Series Analysis does include Trend Analysis (Weekly, Monthly, and Yearly Calculations and Trends), Seasonality Analysis, in the time-series process is to identify the seasonal patterns represented by each month of the year, the graph of monthly call volumes, and trend rate calculations. Monthly forecasting Adjustments, Day-of-week Forecasting, and Time-of-day Forecasting.

Future forecasts using a Forecasting Funnel Approach further break down an annual volume of calls into an hourly or half-hourly estimate:

• Annual Call Volume
• Annual Growth Rate
• Seasonal or Monthly Factors
• Day-of-Week Factors
• Time-of-Day Factors

Funnel steps of Forecasting include:

• Last year's call volume
• Anticipated annual growth rate
• Anticipated annual volume next year
• Percent of annual calls to arrive
• Anticipated Volume
• Number of days the operations are open
• Number of calls for an average day
• Percent of calls in peak hour
• Anticipated call volume during peak hour

The other most important components of the forecasts are the Key Performance Indicators (KPI) including Handletimes, Ring, Queue, Talk, ACW, Shrinkage Factors, etc.

#contactcenterworld