Over several years, bike wheels have been tested for aerodynamic performance. To date, this is the largest independent dataset available. This dataset was originally derived from wind tunnel test data but has latterly moved over to Dinitriev number with a back calculation to give “wattage”. It is not a perfect test.
It should be noted that bicycle wheels are particularly sensitive to tire size and having a wheel and tire combination that bulges will cause a significant increase in drag. Ideally, the tire and the rim should be of the same width. These graphs are now updated on a regular basis and are plotted for interactivity. You can see some of the background for this data here.
Small differences in watts should generally be ignored as these will be affected by arbitrary things such as shoe overlap and the size of bike frame. A bigger differential will produce a bigger drop in drag but it diminishes as the wheel tends towards a disc. In general, a deeper rim will always have less drag. The spoke area which is of high turbulence is much smaller on a deeper wheel – hence the primary drag reduction.
A number of cyclists get infatuated by very small differences between the wheels. Guidance in this regard is to adopt a 2.5SF rule. That is to say that for an appreciable difference, there should be a difference of 2.5 significant figures.
eg a rider is unlikely to notice a difference between a 190W wheel and a 180W wheel (2SF). There is some error in there as well as geometrical differences based on the rider and their bike. When the difference becomes 190W vs 175W, that is effectively 2.5SF and would be noticeable.
This is a rule of thumb and requires considered application.