Beyond the beaten paths of forecasting call center arrivals: on the use of dynamic harmonic regression with predictor variables

Abstract Modern call centers require precise forecasts of and e-mail arrivals to optimize staffing decisions ensure high customer satisfaction through short waiting times the availability qualified agents. In dynamic environment multi-channel contact, organizational decision-makers often rely on robust but simplistic forecasting methods. Although literature indicates that incorporating additional information into time series predictions adds value by improving model performance, extant research in center domain barely considers potential sophisticated multivariate models. Hence, with an extended harmonic regression (DHR) approach, this study proposes a new reliable method for arrivals’ is able capture dynamics include contextual form predictor variables. The evaluates predictive approach arrival leading German online retailer comprising 174 weeks data. analysis involves cross-validation expanding rolling window over 52 comprises established as well machine learning models benchmarks. DHR outperforms compared regard forecast accuracy broad spectrum lead times. This further gives insights selection optimal implementation marketing-relevant variables such catalog releases, mail postal reminders, or billing cycles.