See this recent thread
, among other past threads, for a discussion of this issue. As you'll see, there is no consensus.
In my post in that thread dated 12-15-12 I listed many ways in which the effects of interconnect cables, speaker cables, digital cables, and power cords are dependent on technical characteristics of the components that are being connected, in ways that often don't have much predictability. I described a couple of examples of how a given cable can sometimes even have exactly opposite sonic effects depending on what it is connecting.
The one-brand or "loom" approach essentially ignores all of those component-specific interactions and dependencies, causing the sonic effects resulting from those interactions and dependencies to become arbitrary. While the mixed set approach provides the user with the ability to optimize the selection of each cable to its specific application within the system, to whatever degree the user chooses to invest his or her time and money in doing so.
It seems to me that implicit in the loom approach is the assumption that all of the interactions, dependencies, and technical factors I listed in that post in the other thread, as well as others that could undoubtedly also be identified, are insignificant in relation to effects that are intrinsic to the particular cable designs, and that have no dependency on what is being connected.
IMO it would be a mistake to make that assumption. Although since your question appears to be limited to analog interconnects and speaker cables, your chances of success with a one-brand approach can be expected to be considerably greater than if your selection of digital cables and power cords were also being limited to the same brand.