The overall efficiency could also be computed from Power Input (aka PV power) and Power Output. Most inverters are capable of show those two measurements already computed (with voltage and amps you compute power than compute efficiency).
I think logging DC can tell you when something happen to block light into your panels. While DC readings do swing, the swing under load is somewhat in a ball park. For instance, one of my arrays swings in between 340 and 370 volts under load. This swing depends on the quality/quantity of light hitting panels, load (from inverter), temperature (de-rating panels), etc. But, in a sunny day with few clouds, if a cloud pass between direct sunlight and panels I can see a huge drop outside the ballpark (like 250V for some minutes). In a very very hot day, the swing range shift downs a dozen volts or shift up in a cold winter day (very rare here I must say). What I want to say is that swing in DC volts will not invalidate its readings. On the other hand, monitoring AC voltage is very very stable, here, a 220V AC will vary from 219 to 223 with very few occasions of 217V, all variations inside range proposed by regulatory agencies. No surprise at all.
In my case I am logging DC voltage on standard voltage register and AC voltage on extended register. I did the opposite at first but for me it is meaningful to see string voltage alongside generation/power quantities and AC voltage on the other page at pvoutput. - my 2 cents