There was a paper in Nature, v388, 21 Aug 97, pp 753-755 with a summary on pp 713-714 that described a test using random protuberances. The random did better than the regularly placed roughness, presumably because the regularly placed roughness produced a coherent vortex structure, albeit smaller than the smooth surface structure. Has the added benefit of not needing to be aligned with the flow, though the test setup in the the Nature paper was flow in a channel, not past an object.

The proper way to test something like this would be to use the smooth surface area of your object as the reference area. That way, any effect of the larger gross surface area are factored into the difference in drag coefficient. However, a vendor presenting the data might disagree with what is proper.