Another option is to use PhotoMechanic and shoot raw only, which is what I do. With that strategy the "JPG" is initially the medium quality (but full resolution) JPG embedded by the camera in the raw file. Any edits done in Capture NX2 will embed an almost highest quality JPG in the NEF.
Photomechanic can then either do a straight extract of that JPG or a "Save As" with specified resolution, quality, watermarking, and other niceties.
I find that in most cases the camera embedded JPG is good enough for web sized images. Otherwise I usually edit the image anyway, or occasionally take the time to essentially re-render the embedded JPG with a "Save as" in CNX2 back to the original NEFs (replacing them).
One advantage of using Photomechanic is that it is far faster than ViewNX2 because it is only extracting the existing JPG rather than re-rendering it. On my Core i7 laptop I can render around 150 NEFs with ViewNX2 in about 5 minutes. With Photomechanic an extract takes less than 10 seconds.
In terms of speed verses quality there are various inherent compromises with either method (and also the 3rd method, which is to create Raw + JPG in camera).