analog mux buffer as well
Well let us be realistic. The inportant things of the MUX selection is the ratio of the ON/OFF impedances, the aperture time and the charge injection.
For slow process sampling (like the DC in your case) you may sample reasonably slow, then the aperture time and the charge injection aren't an issue, because you may wait long enough after the channel selection for the system to settle before initiating an ADC conversion. Hence in this case the most important criteria is the ON/OFF impedance ratio. Then how could some put a number on a OPTIMAL one (the most cheap that still do the job) ? Well all depend on the inter-channel xtalk you are allowed to have, and/or of corse on the resolution of your ADC (taking in account eventual oversampling).
Also when measuring DC, you will need good long-therm stability. Since every MUX impedance is thermaly dependent, U should think of implementing differential MUX (preferably on the same die istead of two items), and make sure the system is well thermally insulated from the air convection.
At last (but not least) the proper grounding could be an issue. Try to allways keep the analogue part on a big fat ground plane, separated from the digital ground. Those two usually are tied together @ the ADC.