The first issue is that the accuracy of the onboard computer calcuations are fairly sketchy.....they will give you a rough idea but not that accurate or consistent.
You also can't just go run through one tank of gas and say you got X fuel economy, then install some device and run through a tank of gas and say you now have X fuel economy. Any industry accepted fuel economy test standard requires at least 3 repeatable runs in each configuration and you have to compare two vehicles to each other.
Quick summary:
Take two as identical as possible trucks in stock configuration and run them on the exact same route at the same time, always using the same driver in one truck and the same driver in the other truck. You need a minimum of 3 runs where the fuel consumption ratio between both trucks are all within 2% of each other, and the time it takes the trucks to run the route has to be within 0.5% each time. Theoretically the ratio would be 1:1, meaning both trucks used the same identical amount of fuel. Though it's never identical and thus the reason you need to do this. There is usually at least a couple % difference between them.
Then install the magic device on one of the trucks, leaving the other one stock, and repeat the runs. If the device helps the ratio will now be something like 1.2:1 (basically the difference between the above baseline raito)....meaning the one truck now does 20% better.
Sounds like the TMC / SAE type three fuel test.. it's ok as long as one remembers that it's designed to test products that are supposed to take immediate effect, if there is such a thing.
on the other hand, just as you wouldn't subject radial tires to a test designed to find the limits of bias ply tires, there are technologies that this type of test isn't capable of evaluating
evaluating a single vehicle for mileage improvements isn't a totally flawed proposition if you have an established baseline, somewhat consistent routes, meticulous recordkeeping, and some basic math skills..
it's funny - a guy can dyno his car once, take it home, and throw on a cat back exhaust, and dyno it again sometime later.. if he shows even the slightest improvement,well, it's high fives all around.. but god forbid someone tries a fuel savings device and more than one day passes between installation and reporting the results..if it fails, that's expected..if it shows an improvement, the guy is labeled a delusional nitwit by people who are suddenly experts in fuel chemistry and combustion.
If a vehicle was getting X miles per gallon consistently beforehand, and gets a consistent improvement of X afterwards, then the device works..someone might say " that's only one car, and doesn't prove anything", but the same can be said about the one truck in test you are describing.. nobody will buy anything they don't believe in, no matter what test its subjected to.
that's always the conundrum - everyone believes in dynos, until a technology nobody believes in does well on one..then they demand real world proof because they say dynos aren't accurate..yet they demanded dyno testing to begin with because they don't trust the real world results.. dynos or real world, it's all an educated guess..eventually you have to spend the money and try the thing if you want to solve the problem..
I'm not defending the fitch thingy, just offering random thoughts.. again, for me to take a fuel saving technology seriously, I would want explanation and proof of the how and why it works..and I'd be more jazzed if I could see results over a range of time, conditions, loads, and environments, as opposed to what it did in one test..
