Here's a swing map for 2017:
If one was to characterize these differences, it would be that the most longtime loyal Labour areas either swung Tory or at least very little to Labour, while the areas that haven't been as Labour-friendly or at least haven't been for as long did swing very hard. Even when the Tories were forecast to win very heavily, such a swing difference seemed to be expected.
Many of them are heartlands of White English ethnics, often where culture and immigration are or have been significant vote swingers. The Black Country, the Thames Estuary, the east coast, and more recently there was a very bad conspiracy to hurt children concentrated among some British Pakistanis in South Yorkshire. What I can't explain: Tyne Tees land (formerly known as the North-East Euroregion). The other phenomenon is LD collapse from 2nd or 3rd in a handful of constituencies where they had built a vote among natural Tories.
Why this all happened to help the incumbent Tories in 2017 is another question. My guess is that Ukip did unusually well among those voters in 2015, and they fell back to their "natural" home (since 1970).