Let's consider some linear ideal infinite medium where infinite plane monochromatic wave is propagating. Fluctuations of density cause relative permitivity to fluctuate and proportionally create additional material polarization. As I calculated, if we decide for fluctuations to have volume V then they create in average additional polarization: $$\delta P \sim E(\varepsilon - 1)\sqrt{T/V}$$ It will become a dipole if we multiply it by V and if we take derivative of it and square of it then we'll have intensity of scattered light proportional to volume as it should be. I'm integrating intensity bcs fluctuations are random and doesn't interfere with each other as I understand.

The issue happens when I'm trying to calculate how scattered light interfere or just add up to initial wave. As you can see fluctuations are proportional to field of initial wave, thus scattered light doesn't have phase shift. Then I get problem of scattered light amplifying initial wave.

And nobody in the whole internet calculate how Rayleigh scattered light interact with initial light. Everybody just forgets about initial wave and for the millionth time derive scattering cross section