Analytical support for predicting cost in complex service delivery environments

Within larger and ever-more complex global service delivery networks, forecasting and managing cost becomes an increasing challenge. While manifold interdependencies and transactions between individual delivery centers can still be captured ex post, forecasting the spending and income and expense (I&E) impacts of management decisions becomes tedious. We propose a business-analytics-based approach—drawing on past spending and accounting patterns—to derive service delivery cost forecasts across virtually all elements of the delivery environment in an automated, consistent, and unbiased fashion. Using a set of time-series analyses, as well as multiple regression models, we predict individual spending lines as well as the cost impact accruing from spending lines based on accounting rules. The approach has been developed for a global IT (information technology) service delivery business and has been tested in simulations across several of its units—producing quality predictions able to beat alternative benchmarks.