That's why I love my HP-3456As, no custom parts, nice calibration pots for each range and usually better specs than the average 1-2k$ DMM nowadays.
Apart from that, if I can't calibrate a DMM myself because I don't have the tools, who cares. It just goes uncalibrated, better than getting ripped off. Usually, a good DMM stays within spec for much longer than the suggested calibration interval. And I won't even touch anything without open calibration procedures, so screw you, Rigol and Chinesium crap friends...
Also with 99% of the things I do, the exact value is quite irrelevant. Fast reading/response, high resolution, good linearity&precision, along with good display and user interface are more important for me, than ppm accurate readings.
Like who cares, if the 15V rail is 15.001V or 15.002V, I want to know wether there is noise in mV range.
Or for checking shunt resistors or power Mosfet RDSon, I need a DMM with high resolution in mV range, accuracy isn't that important, as long as it's like 0.1%