I’m still not sure exactly how it works, but the University College London (UCL) has setup a “Surname Profiler” database. Here’s their description:
A recent research project based at University College London (UCL) has investigated the distribution of surnames in Great Britain, both current and historic, in order to understand patterns of regional economic development, population movement and cultural identity. This website allows users to search the databases that we have created, and to trace the geography and history of their family names.
Right now it’s fairly limited (only the years 1881 or 1998), but it has me very intrigued – it shows a distribution of surnames color-coded by area, i.e. you type in “Keith”, and it will show you those areas where the surname “Keith” is most prevalent – in this case, Scotland, and it breaks it down into smaller areas, and you can toggle back and forth between 1881 and 1998, which is, in and of itself, quite interesting. You can see a pseudo-migration pattern.
It could possibly help with research, but it’s more of a stepping off point – i.e. you could play the law of averages and research in those areas where your surname is most likely to be found.