Digitalization of content and exponential growth of Internet and electronic commerce are changing the media industry. The availability of structured content enables new ways to produce and deliver information. This paper explains the role of semantic metadata in developing content for an adaptive news service in the SmartPush -project. In SmartPush, news content is categorized using semi-automatic tools and pre-defined vocabularies. Metadata enhanced content is then matched against user profiles to provide customers with a personalized news service. After providing the personalized news to the customer, SmartPush system adapts the personalization based on user feedback. This paper discusses the requirements of personalized content services and challenges in an approach based on structured metadata. We describe how supporting ontologies for the content were developed and maintained and what kinds of tools were developed to support the structured metadata creation. We also present some results of the pilot phase of the project and introduce some of the issues observed during the system implementation and in the performed field trial.