Evaluating the Utility of Geo-referenced Twitter Data as a Source of Reliable Footfall Insight

Proficient intelligence of footfall is critical for retailers and location based services seeking to exploit the optimum locations and manage existing outlets more efficiently. Sources of data representing daily population movements can provide valuable information on the nature of the temporal demographics across different places. One such source is geo-referenced Twitter data, a sample of which is freely available through Twitter's API. Spatial densities of geo-referenced Tweets can infer the populousness of places at given times or dates. Further, it may be possible to infer the broader travel patterns of users who have visited a specific location by looking up their total activity as recorded in the data. Using a set of heuristics which locate the most likely correct residential addresses, Twitter data can be used to estimate the customer catchments of particular places. It is similarly possible to infer a 'day time catchment', which may be more useful for convenience outlets in proximal locations to places of work. This paper aims to explore the practicality of Twitter data toward retail and service location planners by investigating a year's sample of Tweets transmitted from the British Isles commencing in September 2012. The paper will identify Twitter estimated footfall and customer catchments for a number of locations in London including a selection of major train stations which are the source of many of England's arterial train routes. The estimations will be benchmarked against existing figures which have been privately recorded by other means, and inconsistencies will then be analysed.