There are countless sources of errors in every scientific measurement including those taken for color management. I think the Achilles Heel of all "canned" profiles is that they were not made on your specific printer, with the particular batch of ink then loaded, with the particular batch of media being used. Art media in particular has pretty significant batch variability, especially if it's canvas.
Each of those (and other things like printhead aging) must introduce a certain amount of variability. I have think that will tend to favor measurements taken locally with a merely adequate measuring system, versus those taken elsewhere with a superb measuring system, but not from the exact same printer, ink, and media batches that are needing calibration.
And that may be more applicable to old printers with tired printheads compared to newer examples. If your profiles are shifting blue to magenta, a hot-dog spectrophotometer is beside the point because the data being evaluated doesn't apply to your specific setup, although it can also mean you're just outside the gamut on the blue.
In one scenario, you can print targets on your system, then have somebody with top-notch equipment and skills create spot-on profiles from them that will compensate as fully as possible for all the particular color quirks in your system. Cheaper than a RIP by far, if all you want from the RIP is profiles. There are several zealot-quality profile makers and color management consultants on this forum, and I've heard Andy is one of them.