The test setup
As with most such tests, the basis here is that the wheels are placed in a wind tunnel (the web page has some detailed images of various elements of the setup). The wheel under test appears to be fixed to an inverted fork, on which there are electrical strain gauges - effectively, these are what assesses the aerodynamic drag. The wheels are said to spin in air air flowing along the wind tunnel: it's not clear whether the wheels are driven, whether they spin due to the air flow, or whether this refers to a pivot that allows the wheel to experience air flow at different angles of attack. Drag was measured at wind angles 0 to 35 degrees. Because the more extreme win angles are infrequently experienced on the road, it seems that an average drag value is computed by giving less weight to the more extreme angles, and greater weight to smaller angles. This is indicated by the Gaussian curve in the graph.
The measured forces are translated into power absorbed at a speed to 50km per hour. No explanation of how this was done (or assumptions made) is given.
Pros and Cons
Pros - the test seems unbiased - no indication that it was sponsored by wheel manufacturers.
Cons - it's not clear how rotating the wheels at race pace would affect the drag that was measured. It's possible they did drive the wheels, but the article does not say so. Only front wheels were tested (which makes sense, because the aerodynamics of the rear wheel is far more complex, as airflow is affected by the rider and the bicycle itself). So, no disc wheels were tested. I also don't understand how they decided on the parameters where more extreme wind angles are reduced in importance for the overall power absorption calculation.
The Results
The data are presented as a table - note that wheels are listed alphabetically, rather than by "aero-ness".
Conclusions regarding rim profile and spoke count
The obvious (and expected) conclusions are that deep rims and low spoke counts are good. The best performing wheel is the Zipp 808, with a rim depth (I believe) of around 90mm. This is pretty similar to my trusty Hed Stinger 90 - a wheel not tested here. Also rating well are the carbon four-spokes and trispokes from Xentis and Hed respectively. Of course, many riders find these wheels rather a handful in extremely windy conditions, so ther are swings and roundabouts there.
And finally...
Switching from a "traditional" aero wheel such as a Zipp 440 to the best wheel (on the criteria of this test) should save 3.2W at 50km/h. What does that mean? I suppose if a rider's racing at 250W (is that a reasonable stab?), then this would represent about a 1.25% improvement. I make this 16.5 seconds on a 22 minute '10'. Quite substantial.