I'm unsure if this deserves a new topic, so I will post this here.
I was just casually looking through my Facebook feed and saw a new DxO Mark post. It compares the Sony Zeiss Vario-Tessar T* FE 24-70mm F4 ZA OSS to the Canon EF 24-70mm f/4L IS USM and the Nikon AF-S VR Zoom-Nikkor 24-120mm f/4G ED (link).
I've noticed a few times that certain lens scores do not match up with the actual sub ratings they receive. This latest post takes that a step further. The Sony lens receives a higher score than the Canon, yet in the sub-ratings (sharpness, transmission, etc), the Canon beats it in every category.
Being a trio of zoom lenses though, the rating varies also by focal length (plus of course aperture). DxO Mark's data however also only proves that the Canon copes throughout its range better than the other two. Notice the Canon's graph has less a smaller red zone than the other two here.
Does anyone know what's going on here?
Edit: I also just noticed the Nikon has 1 point more than the Canon, even though it also has a worse score in every measurement!
DXO "scores" should be ignored. DXO scoring is weighted, and how they weight (especially for lenses) is difficult to discern or else completely unknown. It seems rather clear that they have certain brand affinities and brand aversions. DXO seems very averse to Canon, despite the fact that, as you say, their lens measures clearly show that Canon lenses perform exceptionally well and should technically "score" higher.
The general rule of thumb with DXO is to simply ignore the scores, and read the measurements. There is still some danger in that, however, as not all their measurements are actually measured. Many are derived mathematically from actual measures, which implies a certain amount of assumption goes into many of DXO's measurements...an assumption of ideal behavior (which, for anyone familiar with reality, is rarely ever ideal.)
Two key "measures" you should steer clear from are the T-stops for lenses (the way it is used, it rates lenses by absolute transmission, and does not normalize the results for comparison...hence the reason a 50mm f/1.8 beats a 600mm f/4, despite the fact that the latter is a vastly superior lens.) The other is Print DR for sensors, as even though it is called a measure, it is not. It is a weighted derivation based on the actual underlying DR measurement: Screen DR.
Last, beware that DXO has been known to change their scoring mechanisms behind the scenes without being clear why, when, or how...so sometimes information changes without prior disclosure, and god only knows why.
DXO information needs to be taken with a healthy dose of salt. It isn't always reliable.