Matt Zook and colleagues used the term back in 2004 in order to describe the ways that personal information can be incorporated into surveillance regimes.
More recently, I've found the term useful to articulate the layers of digital information created about places. In other words, the ways in which various contours of places can be represented on our digital canvases. Data shadows can take a variety of forms, but most are manifestations or byproducts of human/machine interactions in what Martin Dodge and Rob Kitchin call code/spaces and coded spaces.
This is all discussed in much more detail in the following publication:
Graham, M. 2013. The Virtual Dimension. In Global City Challenges: debating a concept, improving the practice. eds. M. Acuto and W. Steele. London: Palgrave. (in press).
Alternatively, the idea is discussed in a somewhat different form in my earlier paper on palimpsests:
Graham, M. 2010. Neogeography and the Palimpsests of Place. Tijdschrift voor Economische en Sociale Geografie. 101(4), 422-436.
The 'palimpsest' notion, however, focuses largely on the layers of information over places rather than the ways that they reflect lived patterns and processes. More recently, Matt Zook, Andrew Boulton, and I have instead turned to the notion of 'augmented realities' as a way of paying attention to the indeterminate, unstable, context dependent and multiple realities brought into being through the subjective coming-togethers in time and space of material and virtual experience. In other words, the ways that everyday life is increasingly experienced in conjunction with, and produced by, digital and coded information. This is all described more in these two papers:
Graham, M., M. Zook., and A. Boulton. 2013. Augmented Reality in the Urban Environment: contested content and the duplicity of code. Transactions of the Institute of British Geographers.38(3), 464-479.
Graham, M and M. Zook. 2013. Augmented Realities and Uneven Geographies: Exploring the Geo-linguistic Contours of the Web. Environment and Planning A 45(1) 77-99.
I'm particularly interested in tracing the 'data shadows' idea further back in time. So would welcome any links and references that help me do that. Comments and thoughts on whether 'data shadows' even serves as a useful function would also be welcome.