The 411 on accurate forecasting

Updated: October 13, 2010

Forecasting is the proverbial mix of art and science. It begins with predicting how many transactions you are going to get in the future. To do that, you look at historical data to determine patterns that reflect when customers contact, and consider possible trends that will effect these patterns of contacting a customer service. Next, we factor the handling times of these transactions. Finally, you modify results based on conditions not reflected in historical data.

In most contact centre operations, long term forecasts look out a year and beyond. We use them to estimate future annual budgets, establish hiring plans and define future system needs.

The short term forecast predict workloads up to three(3) months. They are important for organizing and adjusting schedule requirements, anticipating seasonal staffing needs, planning for holidays, and determine imminent hiring requirements.

Weekly, daily, and intra-day forecasts are called short term tactical forecast and are used to tighten up schedules and adjust priorities around current conditions and near term events.

Five Key Activities

The basic historical data you need for forecasting includes how many inbound transactions you have received in the past, when they arrived and how long they took to handle. Five key terms reflect this activity:

Talk time: everything from the beginning to the end of the call, and in between. Also ATT (Average Talk Time).

After call work (ACW) or Wrap-up time: time that is needed to log a transaction or to log data into a customer information system.

Average Handling Time (AHT): average talk time + after call work.

Call Load: the volume of transactions coupled with how long they last. volume multiplied by average handling time.

Abandoned calls and busy signals: customers that hang up after a period of time before talking to a CSR are registered as abandoned calls. When a customer is getting a busy signal, it means all lines are busy and the IVR is overloading.

If the historical data used to forecast ignores ones of the above activities, it will be underestimating demand.

Just remember, the forecast should always reflect "offered" transactions. In other words, offered transactions are all of the attempts your customers make to reach you. We have several possibilities, such as answered by a CSR, answered by the system (IVR) but hang up before reaching a CSR, or they get a busy signal.