At match lengths where the results are indisputable, and I generally take anything over 20 cM as falling into that category, the differences are minimal and not worth bothering about IMO. Yes, one test might give you a slightly larger total cM than another, but how do you know which test is the more accurate, if either?
One caveat is that if you have tested with a company that provide segment and chromosome data, I can see the value of comparing tests taken with the same company.
For smaller matches, where the chances of the match being false (IBC etc.) are significant, I don't regard a match alone as having any significant value, unless it is corroborated by other research, in which case each piece of evidence has a value in adding weight to the other.
As I mentioned above, if you create a superkit on GEDmatch from two or more tests taken with different providers, you will lose quite a few of the smaller matches, because with more actual data to compare and less imputation, some of the false matches are stripped out.
Unfortunately, GEDmatch has a much smaller database than some of the other companies, particularly Ancestry, so again the value is limited unless a potential match already has their test on GEDmatch or they can be persuaded to upload it for comparison.
