Market research companies spend large amounts of money carrying out time-intensive processes to gather information about peo- ple’s activities, such as the place they frequent and the activities in which they partake. Due to high costs and logistical difficulties, an automated approach to this practice is needed. In this work we present an automated market research system based on computer vision and machine learning algorithms with visual lifelogging data, developed in collaboration with Sponge It, a market research com- pany. Due to some image quality constraints associated with the Sense- cam, for our prototype system we developed a visual lifelogging device using an Android smartphone. This device can capture images at higher resolutions and with additional metadata, such as location information. The aim of this project is to analyse large collections of visual lifelogs and to support both ethnographic research and audience measurement for market research. Ethnographic research is supported by high level classification of images to capture the semantics of the users activities (e.g. socialising in bar, shopping, eating). Location, time and other con- texts are also analysed, and an interactive interface supports browsing and exploration of the data based on this analysis. The system can measure audience exposure to specific advertising cam- paigns, using object recognition algorithms to automatically detect the presence of known logos in life logging images. This combination of con- cept classification for ethnographic research and object recognition for audience exposure represents a very powerful tool from a market research perspective.