Demand forecasting for Customer Call Centres

Customers interact with organisations via multiple channels. But whilst cost efficiency provides a clear motivation for businesses to increase customer self-service, many banking clients still prefer to call and talk to an agent.

BJSS’ client, a top-3 retail banking group, dedicated 80 full-time staff to manually forecast the likely demand on its network of call centres. This would inform the number of customer service staff needed at any half-hour interval to answer customer calls.

It was a manual, subjective and heavily error prone process.

The existing manual process relied heavily on subject matter expertise, and deep personal experience rather than systemised knowledge. Therefore, the most immediate need to impact on performance was to improve efficiency by automating this forecasting process.

BJSS built the client’s demand forecasting tool to drive optimal staff resourcing levels.

Accurate forecasting is critical to successfully managing a call centre. To make sure that there are enough agents to attend to clients, organisations deploy methods to precisely predict the call volumes they are likely to face. However, predicting the “future” is challenging and inaccurate forecasts can result in under or over-staffing, resulting in a poor customer experience or operational inefficiency.

THE BJSS SOLUTION USES PYTHON AND PROPHET AND INCLUDES A FRONT-END REBUILD AND DATA PIPELINE AUTOMATION.

The BJSS solution has delivered time series analytics that include seasonality, covariate handling for events, and structural breaks.

This highly flexible, scalable solution has allowed the client to automatically produce 18-month forecasts in 30-minute intervals. It has also reduced forecasting errors by 15% – improving the client’s resourcing decisions and customer satisfaction.

However, this is just the starting point. On a larger scale, the client has engaged BJSS to deliver a modernisation initiative that covers many of its customer and non-customer facing capabilities.