Horizontal CMM Experience? Looking for info...

Two questions:
Did you verify that your CMM program was outputting good numbers by checking some dimensions through other inspection means?
Are all three of you using the exact same program?

If yes and yes...and if 2 out of 3 are saying its good, and the 1 saying they are having issues hasn't been calibrated in 5 years...... then I think we have found our issue.

To Milicron Mike 's point....I hate horizontal CMMs due to the exact issues he has laid out. There is a fine science that must be learned & applied in order to measure properly and accurately.

Oh boy. "Pizzing" contests aren't good (especially if they're people you're trying to get money from).

I would do is this:
1) Get a copy of your CMM calibration cert. Hopefully you're fully calibrated to a NIST standard. This proves your machine is calibrated & working properly.
2) Write a quick little program to measure something like a tightly toleranced thread ring THAT HAS BEEN CALIBRATED TO A NIST TRACEABLE STANDARD. I'd do it 30 times...each time re-aligning the part before DCC measurement. Collect the data. Run through CPK formula to provide further data that your CMM is repeating on this gage. Once/if satisfied....
3) Request for your fixture back. Measure it 30 times...each time re-aligning the part before DCC measurement. Collect the data. Run through CPK formula to provide further data that your CMM is repeating on your fixture.
4) Last but not least....do a 100% layout inspection of your fixture on the plate. Collect the data in an inspection report. Hopefully all checks out well.
5) Send them: your CMM cal cert, CPK data on your ring gage or whatever you choose, cpk data on your fixture, and accepted layout inspection report.
6) Ask them to prove to you respectfully yet objectively how they can reject two reputable source's data with an un-calibrated piece of junk (i suggest you choose your own working but I'll leave that up to you).

Oh boy. "Pizzing" contests aren't good (especially if they're people you're trying to get money from).

I would do is this:
1) Get a copy of your CMM calibration cert. Hopefully you're fully calibrated to a NIST standard. This proves your machine is calibrated & working properly.
2) Write a quick little program to measure something like a tightly toleranced thread ring THAT HAS BEEN CALIBRATED TO A NIST TRACEABLE STANDARD. I'd do it 30 times...each time re-aligning the part before DCC measurement. Collect the data. Run through CPK formula to provide further data that your CMM is repeating on this gage. Once/if satisfied....
3) Request for your fixture back. Measure it 30 times...each time re-aligning the part before DCC measurement. Collect the data. Run through CPK formula to provide further data that your CMM is repeating on your fixture.
4) Last but not least....do a 100% layout inspection of your fixture on the plate. Collect the data in an inspection report. Hopefully all checks out well.
5) Send them: your CMM cal cert, CPK data on your ring gage or whatever you choose, cpk data on your fixture, and accepted layout inspection report.
6) Ask them to prove to you respectfully yet objectively how they can reject two reputable source's data with an un-calibrated piece of junk (i suggest you choose your own working but I'll leave that up to you).