So the trick is in knowing your camera and its practical DR. The measurements on dxomark.com may be scientifically accurate but I think most people would agree that they are over-optimistic and about two stops too high. For my 5D2, shooting Raw, I figure about 10 stops for ISO 100/200, 9.5 stops at ISO 400, 9 stops at ISO 800 and around 7.3 stops at ISO 1600. Above 1600 subtract a full stop for each doubling of ISO.
Well I wouldn't say they are optimistic since they never say you will be able to capture the DR figures they provide. They just took an engineering definition of DR based on a SNR=0dB.
For practical photographic applications, a much more logical criteria IMO is SNR=12dB, which is a good representation of the max amount of tolerable noise in general photography:
Since DR is defined by the SNR in the deep shadows (which is where the presence of noise really hace consequences), and the constant read noise is dominant there over photon noise, the SNR improves by 1 entire stop (or 6dB) for every extra stop of RAW exposure. That means that we can take DxOMark's DR figures and correct them by -2 stops of practical DR, which is the same correction as you suggest.
So in practice: want to know the DR of real world scenes you'll be able to capture with a given camera at a given ISO without needing to make several exposures? take DxOMark's DR figure, substract 2 stops to it, and that's it.