Dear Arpid! You know that backscattering light (BL) in RLG is the reason of coupling between the clockwise and counter-clockwise waves when the rotation velocity of Gyro is small to provide the enough gross Doppler shift between them. The coupling stopts functioning of Gyro to measure  the rotation velocity. That is why we need to make the backscattering light as small as possible. Main source of origin of BL is reflection on mirrors in the case  of  gas He-Ne laser or  BS in  waveguides. In linear optics approximation BS coefficients stay permanent  while BS intensity varies if lasing power changes. BS coefficient can changes in time with aging  all surfaces responded for BS in Gyro. So that is  my anwer  on Your question.

All the best,  

But what about the phase of the backscattered waves. It is not constant and thus each backscattered wave adds randomly to other. This may make the overall backscattered wave to behave as a random wave.

I see and accept your arguments..

My doubt is - Is it possible to quantify this conservative backscattering phenomenon ?

Some information on this question may be found, for example, in the work:

E.A.Bondarenko. A laser gyro with a four-mirror square resonator: formulas for simulating the dynamics of the synchronisation zone parameters of the frequencies of counterpropagating waves during the device operation in the self-heating regime. Quantum Electronics 44, 364 (2014).