You may be in for a challenge as the process and progression you are using could be a major factor in the poor fusion.
Even with the most trained welders downhill GMAW can be a challenge to achieve good penetration and fusion. The desire to slow down to "hold" the puddle size can be corrected by increasing travel speed however depending upon electrical settings you may be pushing the limit on too little heat input. Too slow and too fast BOTH result in fusion issues. One because the arc doesn't melt the underlying base metal due to energy being wasted heating the puddle and the other because not enough energy is used to melt the base metal because of high travel speeds.
The lack of fusion can be minimized by process controls however the process you are using is more prone than most others to this type of defect. If the welds are off center then a good visual inspection program and production planning may be another aid. If the welds are so restricted that the welders cannot see the joint then either they need more practice or the assembly sequence may need to be modified.
Proof testing a structure would be diffficult. After the proof testing you may have caused MARGINAL welds to fail but they may not be visible to the naked eye. Then the product gets to go in service and fails even faster.
Welding is considers a "Special Process" and is difficult to verify the integrity of the finished product by many methods. Selecting a more forgiving process, additional welder training, and review of the design would be a better option in my opinion.
Can you give some more specifics on the issues you are having . Material thicknesses, Voltage and WFS/AMPS, joint designs, base metal cleaning (mill scale removed) etc.
PLease forgive me if all of the above are things you have taken into account. The GMAW process like all others has advantages and disadvantages over other processes. Its easy to make a good looking downhill weld that can be removed with a claw hammer.
Have a nice day.
Gerald Austin
www.weldingdata.com