Footloose information technology producer services are increasingly viewed as the new engines of regional growth and employment creation. The claim is often that appropriate government policy can engineer comparative advantage in that direction by means of judicious subsidisation. The evaluation of such claims relies on accurate statistical data. We argue that attempts to distinguish in survey data or official statistics the subset of services commonly (but unofficially) known as producer services from consumer and government services may result in a potential 'bias' with implications for public policy. An input-output model is developed for estimating this bias by decomposing service industry data into its producer and consumer services content. This method can give a truer estimate of the actual employment in producer services at various spatial levels. We find a systematic bias in official data and a non-systematic bias in survey data. The model is applicable to countries and regions for which input-output data are available.