What exactly are you arguing for?I'm arguing that lamenting "mechanization" of social interaction - whether between corporate employee and client as in this example, or even between individuals - is akin to appealing to ether as a virtue of the universe.
If interactions between actors can be measured (which they can to some level of specificity in certain contexts), and the desires of the actors can be understood to such a degree (which we're starting to be able to do), then we can model actions which increase or reduce the likelihood of desired outcome, and following from that can produce and optimize decision support systems that nudge users toward some mutually beneficial state that may have otherwise been opaque to both actors.
People seem to hate this idea because it basically puts hard determinism right in their face - in that your past behaviors if known well enough should be predictive of future actions ceterus paribus.
It's often argued that, such a granularity of measurement in social dimensions is impossible technically, or the fact of being measured changes their behavior. I wouldn't argue either of those to be untrue, only that it need not be perfect to increase the overall optimization of the system.
So instead of saying, yes lets use measurement to optimize our system of interactions across commerce and relations, people bristle at the mere concept of social engineering in the Popperian sense of the term because it feels restrictive to our sense of "free will." I argue the opposite, that doing such would simply make us more aware of our predilections and much more likely to be able to align them across groups.