I'm not a big fan of DXOMark but at the end of the day I'll peruse all available review sites when making a buying decision. So when looking at their lens reviews, I noticed that they give the Canon 85 1.8 the HIGHEST resolution score of ANY lens they've reviewed...including nikon, zeiss, sigma...etc. I know it's a very good lens but I find this hard to believe.
One-size-fits-all methods for scoring lenses are doomed to fail because the different lenses are all inherently different, have different design tradeoffs, purposes, etc. For example, which is "better", a macro lens or portrait lens ? (usually, the macro lens is better for macro and the portrait lens is better for portraits)
Besides differing constraints (e.g. a travel zoom need not have a fast aperture, and a macro lens doesn't need a fast aperture), lenses of different focal lengths aren't comparable.
So for a review site to be useful, the review needs to do a good job at presenting and summarizing the measurements, including at least some discussion of subjective factors or factors that are otherwise not as easy to measure (bokeh, usability, AF performance), and putting it in context (e.g. how does the lens do against its peers ?). Sites like thedigitalpicture, photozone and lenstip do a pretty good job at this.
I haven't paid much attention to their scores for lenses because the other sources do a better job at reviewing them.
Their sensor reviews are quite good but their lens reviews aren't as useful as their competitors.