Most fuel flow gauges are pressure sensing devices.
If you saw something like
this then that would use a remotely mounted turbine transducer, typically a Floscan 201B, which sends pulses back (of the order of 30,000 per USG) which are totalised. Such systems can be accurate to 1% or better, if they are correctly installed and calibrated. All TB20s made after 2000 or so had their Shadin totaliser out of cal by anything up to 25%; there is a U.S. STC fix for it but it's not legal to implement it on a G-reg
Dublinpilot's "tacho unit" method is interesting. I guess this tacho simply counts engine revolutions which, in typical usage and with a fixed pitch prop, is going to approximate total fuel used. It should not be that close however because prop thrust is the cube of RPM, so fuel flow should also be the cube of RPM, whereas the tacho is just counting straight RPM.