I view DXOMark as something of academic interest, but don't usually include it in the websites I consult prior to any lens purchase. It's not that useful on its own, particularly for a less experienced photographer who might end up obsessing over the differences in a particular metric between two lenses, when it has no practical bearing on real-world applications, such as rejecting a portrait lens contender for poor corner sharpness, or a lens you mean to use for landscapes for being a bit soft wide open.
The OP is right to look at examples on Flickr and 500px. What people are actually doing with a lens is a far better measure of its performance. After that, reviews from the-digital-picture, photozone.de, digitalrev, tony northrup, fro, etc... all usually have some interesting bits that go beyond the specs and end results and consider things like build, ease-of-use, real-world gotchas, etc. that you don't get from numbers and charts.