Earlier quoted context omitted.
> From my personal experience most times the hardest part is to convince the users. Former actual LiMux team member here, so I can speak from experience (almost a decade ago though). The hardest part was convince users to be happy on utterly inadequate and aged hardware . No, 1 GB of RAM won't cut it when you have to have Firefox, Thunderbird, OpenOffice and some fucking Java crap open at the same time, and 1024x768…
Yeah, that must have been a huge problem. These days however an OKayish notebook can be had for $400 or an i3 NUC with 8GB RAM and 256GB SSD with an IPS 1080p monitor for $500, and those would run any Linux well for foreseeable future without killing city's budget.
The last part alone excludes many cheap vendors as they don't have government contracting experience, and the spare parts requirements kills the rest as they can only be so cheap because they don't have to build much of a spare parts infrastructure. (Note: this is not specifically relating to LiMux in any way, I have no knowledge about the contracts/tender process that happened back then)
Also, the scale matters: 15.500 PCs are even at 500$ apiece 7.75 million €.