Fuel efficiency is VERY hard to accurately measure in the field with regard to input changes; it's much more easily studied in the lab where the variables are tightly controlled. I doubt anyone can truly prove any substantial increase in a few runs to/from somewhere.
The point of the PEA is to "clean" the injection system. (mainly improve the injectors to a state of being like new in their spray pattern.) Claiming a 7% increase in efficiency would indicate to me that the PEA "cleaned" a very dirty system; a bold claim purported to happen in one tank's worth of application.
But, once the system is "clean" from one or two applications, there should be no "improvement" thereafter ... (because you can't clean something that's already clean). Hence, where one could claim (unsubstantiated) that a 7% improvement is possible, it would be improbable to duplicate upon continued use. And to claim it's repeatable in a system which has already been "cleaned' is, IMO, absurd.
I do believe that PEA can do some really good work when heavily dosed; there was a Gumout demo held for BITOG about a decade or so ago; I believe wwillson attended that event. But I don't recall all the dets. I think they did a visual inspection and the cylinders/pistons were much cleaner after application and an overnight full-on run of the vehicle. However, I'm not sure if they did any fuel efficiency assessments.