We explore gravitational-wave transient cases from the perspective of the two-body problem in curved spacetime, starting with the first case, GW150914, which marks the GW discovery [1]. In this paper, the LVC authors estimated the characteristic (chirp) mass of the binary blackhole system emitted this signal. Their calculation was based on Numerical-Relativity (NR) templates and presumably accounted fully for the non-linearity of GR. The same team later presented an alternative analysis of GW150914 [2], using the quadrupole post-Newtonian (PN) approximation of GR. Both analyses gave similar results, despite being based on quite different assumptions about the linearity or non-linearity of the coordinate reference frame near the GW source. Here we revisit the PN-analysis of GW150914. As in paper [2], our result also agrees with the NR-based chirp mass value published in [1]. Additionally, we apply the 
PN-approximation formalism to the rest of the GWTC cases, finding that practically all of their PN-approximated chirp masses coincide with the published NR-based values from GWTC. In our view, this implies that the NR-based theory, which is currently in use for processing GW signals, does not fully account for the difference between the source and detector reference frames because the PN-approximation, which is used for the comparison, does not account for this difference by design, given the flat-spacetime initial assumptions of this approximation. We find that the basis of this issue lies in the missing information about the derivatives of GW frequencies in the source reference frame. This leads to a systematic error in the estimated chirp masses of GW sources. The corresponding luminosity distances of these sources also turn out to be overestimated.