Many thanks Dude and Lavoisier.
My thoughts are that many reactions could be monitored by triangulating with two analytical techniques. The second technique doesn't have to be calorimetry (although the advantage of calorimetry is that you don't need to know what the impurities are in order to measure them).
Take for example a reaction like this:
(1) A + B -> AB
Also
(2) A + B ->ABB
If you measured this reaction with NIR (true yield) and observed a low yield of AB it could be due to a slow reaction or it could be due to the conditions favouring the formation of ABB (or something else). If you also monitored the reaction with calorimetry however, this would tell you whether the reaction was going slowly or that another reaction was going on at the same time (since apparant yield - true yield = other reaction). If you can make this distinction in real time, you can determine optimum conditions (pH, temperature, addition rate, agitation rate, pressure, accumulation etc) in a single experiment. Using a real time method like this might also give a better insight into processes where transient effects are of interest (e.g. was the yield low because the method was poor or because something else happened later). I suspect that this approach might also work with physical process change like crystallisation.
Are the experiments (acetophenone, HBr/1,3-butadiene) you quoted common in real life chemistry? I am keen to find an application of real practical value.