The geodesics is, of course, only an idealization - for a pointlike test particle with negligible mass. Not that it is not accurate. To compute the trajectory of the Mercury around the Sun it was accurate enough to compute the perihelion shift. But it would of course fail for the computation of a nearby satellite, who would be influenced by the mass of Mercury too. Which the gravitatational field of the Sun, with Mercuries geodesic trajectory on it, would not describe.
The velocities of the binary pulsars, which have given via their energy loss the first implicit evidence of gravitational waves, are, of course, not small.
https://en.wikipedia.org/wiki/Hulse–Taylor_binary gives a 7.5 hours orbit period and a size at maximum of 3,153,600 km. So the light needs 10 sec. for this. 7.5 hour for what light needs $\pi$ times 10 sec is not what I would name a "meaningful percentage".