Along with ROI there are many reasons to have on-site lab
AT NORIA conference this past week - most heavily attended session was probably on sludge and varnish due to problem power generation is having with varnish in its turbines. Due to changes in additives and basestocks, the common techniques such as TAN, viscosity and particle count as well as RPVOT and FTIR (if amines used in antioxidant package) have proven to be unreliable in predicting varnish - one trip caused by varnish is estimated to cost over $75,000. So an on-site lab with equipment specifically designed to predict varnish is easily justified.
US Air Force only uses emission spectrometer to monitor wear metals - need to have analysis done after each flight - no time to send out for oil analysis - high top off rates keep oil fresh so currently no need for oil condition analysis. However, tighter/more powerful engines will make oil condition analysis relevant - turn around time will require on-site
Our Instron lab was losing one design of hydraulic pumps prior to warranty period even though oil condition OK. Determined ZDDP was depleting followed by failure so on-site analysis combined with routine off-site used to monitor that specific failure mechanism as well as over all system health
So on-site could be due to changing oil formulations, changing equipment designs, fast turn arounds, specific failure mechanisms as well as ROI.