I only made the comment because you discounted empirical/controlled comparisons between flagship phones. Without these type of controlled comparisons, EVERYTHING is anecdotal. We all do things slightly differently... and no doubt if you owned 3-4 flagship phones (S6, M9, LG G4, etc...), you'd probably have similar apps installed, but you probably wouldn't have them EXACTLY setup the same, nor have the Android settings set exactly the same. In which even your views on battery life between them would still be considered anecdotal.
These guys at Anandtech, GSMArena, etc... aren't fools. They know that that when they run these battery run-down tests everything must be set identical, and they run tests that simulate real-world use as best they can. Video run-down, web browsing simulators, etc.. So it just doesn't sit well with me when people discount these tests, since it doesn't 'simulate' how they use their phone. We all can 'take them with a grain of salt' if we set our displays to 0-10%, turn off all data sync from every APP, turn off animation acceleration, and then magically claim that 'I get 5-6 hours SoT!! Those dudes at 'INSERT TECHSITE' are whack!'
But I agree with you.... can't hurt to try out a phone and returning it if it doesn't meet your needs. For those that don't want to go through the hassle though, I think reviews from the 'trusted' sites are invaluable.