those guys are getting a lot of mileage out of their "cfd" marketing claims.If you follow the diy hypermile guys you'll see all kinds of tail extensions. Also unplugged performance did a cfd on the spoiler and showed a gain in efficiency.
cfd means they plugged a model into a computer. the cfd is only as useful as the accuracy and of the model, the simulation engine, the rigour of the configuration, etc. the simulation spits out a lot of noise and uncertainty, so in any industry ever of any consequence that uses cfd whether its f1 racing cars or even as 'basic' as bicycles they spend $$$ to put the models into a wind tunnel, measure the "real life" results, and use that to calibrate and recalibrate the cfd "computer model".
i work with this stuff as applied to a field where a single company is larger than the entire automotive aftermarket industry, and i can churn out fancy looking simulation results with multicolored cells and squiggly streamlines...and it can be entirely fictitious. so no i dont put any credence on a private, small business in california, making any claims on aerodynamics on the basis of 'cfd'.
british statistician coined a popular phrase used in this practice, "all models are wrong, but some are useful"