Contact Center Series: Contact Center Forecasting and Scheduling
By Ken Ng

Contact Center forecasting and scheduling ranks as one of the most misunderstood disciplines in the Contact Center industry. I have observed operations in utter chaos and in the worst case scenario, grind to a halt from under or over utilization of the staff and attrition rates. This article will attempt to clarify the proper approaches and importance of precise forecasting and scheduling.

Although operational costs vary from one Contact Center to another, staffing costs inevitably always account for the largest expenditure in a Contact Center budget. Some Malaysian Contact Centers experience up to 70% of their budget allocated to staffing costs. Therefore, staffing is one resource that requires maximum utilization to realize a positive ROI. If the contact center manager is able to maximize the use of the people in the contact center the impact on the budget will be considerable.

The key to unlock the mystery of forecasting and scheduling lies in precise call forecasting. This is important because it enables you to determine exactly how many staffs you will need in the contact center at any given point in time. Overstaffing your contact center will translate into high costs, resulting in under utilization of headcount. You will find staff sitting around surfing the net or reading magazines while waiting for the next customer contact to arrive in their respective call-masters. However, understaffing could be more detrimental to the overall health of your contact center. The outcomes could possibly be long waiting times for customers with the repercussions of taking a drastic hit on your Customer Experience Index (CEI) and loss of revenue if you are in a sales driven operation. Most damaging of all, dissatisfied and overworked staff will contribute to a high attrition rate.

Forecasts can be looked at in three durations; short-term (up to three months), medium- term (from three months to a year) and long-term (up to two years). If you run an established contact center, you will need to forecast yearly for budget allocation and it is wise to generate periodic short-term forecasts throughout the year as the environment changes.

Kevin Hook in his book The Human Face of Call Centers describes two elements to consider when generating periodic forecasts of call volume for your contact center:

1. The total number of calls that will arrive over a specific given time period (three months, a year etc.)

2. The way in which on a monthly, weekly, daily and hourly basis you are able to identify trends in the pattern of contacts arriving

Your basis for periodic call volume forecasting is historical data. Assuming your contact center has been in operation for some duration of time, any decent ACD will be able to provide you with the appropriate historical call volume data that you request for.
Assuming we are looking at element two as explained above, an analysis of your historical call volume over the last year or two will yield findings of your calls arriving in patterns depending on the environment or situation. These patterns, as you will go on to discover often time repeats itself; the hours between 9:00 a.m. and 11:00 a.m. receives a high call volume, Monday is a busy day, December is a quiet month or the festive seasons see your contact center assaulted with high call volumes. A high level of perceptiveness and acuteness to specific factors need to be in place when analyzing the call volume data e.g. non-working Saturdays see an increase in calls as opposed to working Saturdays, the knowledge of marketing running campaigns during a certain period of time always impacts call arrival, the last week of the month where your organization mails billings out, call volumes drop during the Olympics.

These trends can be identified and factored into your forecast as a percentage deviation from the average call volume for the month, week, day or hour. So, for example:

Unless you can accurately forecast the total volume, all this is meaningless. There are several statistical techniques for forecasting which fall into two categories:

Time-series forecasting
Projects past patterns and trends into the future using rolling averages, and it works on the basis that established patterns will be repeated and are reasonably accurate for short term forecasting (most forecasting software packages for ACDs use some kind of time- series analysis)

Explanatory forecasting
Techniques such as regression analysis seek to reveal the links between sets of variables to produce a forecast, for example to establish the link between past changes in pricing structure and call volumes in order to predict the impact of future changes.

In a customer service environment the number of contacts made is linked to the total number of customers, the more customers you have the more calls you are likely to receive. Therefore it is possible to forecast contact volumes by establishing the historical relationship between the size of the customer base and the number of contacts received.

Last updated - 31 August 2004