A full NC-Checker machine benchmark on a new machine after the vendor thinks it is setup is a fantastic way of ironing out these issues. Example issues include the head, or rotary table rotation, axis parallelism or tool loader. All be revealed in a tabulated hard copy form for evidence, as well as many more settings and measurements.
The advantage for the vendor is that the engineer is at your site, he is next to the machine and the changes he needs to make usually take seconds. Doing this the traditional way involves months of work machining with a sense that all is not well, weeks of argument with the vendor, organising a visit for a random overhaul – because nobody really knows what the problem is, even you. And the vendor engineer on-site for long periods of time.