The introduction of FLEX take-offs on the AVRO-RJ fleet in my previous company (about 5 years ago) led to a fierce debate amongst pilots.
Chiefs were telling us to flex as much as possible to save mucha €€, we pilots could not believe that we had to increase power after "thrust reduction"...
Being an unexperienced pilot back then, and after 5 years of asking my collegues about how they operate, I finally made up my own mind.
1) FLEX does save maintenance costs
2) FLEX does burn more fuel
I made some calculations and actually, the extra fuel burn would offset the maintenance costs by 50% on the RJ. That was a few years ago with a barrel below 50 $ IIRC.
3) Performing a FLEX take-off at an average gross weight on a long runway gave us as much margin as a a REF take-off at max take-off weight on a long runway. In both cases we would not take-off in balanced field conditions, so plenty of margin.
4) Performing a FLEX take-off in balanced field conditions (short runway + average gross weight or long runway + high gross weight + high temps for isntance) is just as tricky as a REF take-off at max performance take-off weight on a medium or short runway (balanced field and/or WAT limited). You have to remind that there is no margin left, you HAVE to perform a static take-off from the calculated take-off position (including the calculated line-up position LUF, LUZ, LUH) and you better make sure every other parameter that was used for take-off is met on the runway (actual OAT on a hot runway vs the ATIS OAT, actual wind component, actual gross weight maybe different than the one you used for calculations, etc etc)
The conclusion of 4) is that in fact, because you are performing a FLEX take-off, you no dot realise that you are in the same position as a REF take-off. Performing a REF take-off, your attention is probably going to be at a higher level. Which means that you might tend to forget about the static take-off, etc.
My personal conclusion is that I used to visualize my margin. For instance, if we were to take off from full length, I would calculate from an intersection. That was going to be my margin.
I also used to increase my assumed weight by two tons (5% for an RJ) so that if I ever was in a situation were we were balanced field or WAT limited, hence a decrease in assumed temperature, we would still have that margin to play with.
Furthermore, when the assumed temp would start decreasing for the actual weight, I used to calculate a "REDU" take-off power, which in the RJ is a 10% derate from REF.
That way everybody was happy, so was I, and we always new how big our margins were in case we needed them.