how it works
The EE coverage maps are based on predicted coverage data from network planning tools, i.e. the tools that the network engineers use to plan the network. The predicted coverage is of course only an estimate (based on propogation algorithms) using simple clutter data, i.e. indicating rural/urban/water/terrain height etc It does not go down to individual building/tree level. This data for 2g, 3g and 4g for both Tmobile & Orange is then brought together.
The important thing to consider at this point, is that marketing choose the signal level which they think maps to good/bad/none for both outside and inside coverage. Their opinion may differ to the engineers ;)
Typically this data is refreshed weekly, although operational issues might mean a longer period between updates.
A massive programme of site reduction was carried out by EE leading to some locations losing coverage, but that's always been the case (i.e. some site leases cannot be renewed, or cheaper sites become available, or config changes are made thus making some sites redundant). Additionally some sites might have been removed as a site lease ended, knowing that a 4g site will be replacing it at some point in the future. The outcome of this constant shuffling of sites, is that some people win and some people lose.
I think the real problem in this debate, is that peoples expectations are 100% coverage, 100% of the time. It's just not possible unless people are willing to pay significantly more to support the massive increase in infrastructure which is required to support it.
Finally, I suggest you take AlbertH's 'evidence' with a pinch of salt, the few people on here telling him he's wrong seem to work in the mobile industry. AlbertH comes across as a serial fantasist (see some of his other posts, I don't know how he manages to find time to post here considering he's making thousands of pounds suing various people).