This isn't personal experience, but some tests were conducted back in the 80s when deflectors like this were popular in Europe - these were proper full scale wind tunnel tests by aerodynamicists, rather than claims by manufacturers.
These drawings show the actual profiles of the three test cases - you may recognise the outline of the then Volkswagen Passat car:
In the first and last cases, the drag coefficient is 0.53 while in the middle case with a deflector it is 0.45 - that's a solid 15% drag reduction and so you would expect a fuel economy improvement of at least 10%.
The other part of these tests was to measure the aerodynamic lift on the front of the trailer, as a reduction in hitch weight
at 50mph. In the first case, the hitch weight
reduced by 75 pounds which, compared to a typical low European hitch weight
(typically 100-150 pounds), is very significant.
With the roof deflector, the reduction was 70 pounds, but with a slope-fronted trailer and no deflector, there was no reduction in hitch weight and so no reduction in trailer stability - this is the main reason why you will find a sloping front on just about every European trailer made since then.
Of course these results don't compare directly to typical American towing practice, but it does show categorically that deflectors can work well in the right conditions.