Sky-averaged 95% upper limit on the GW strain amplitude from a circular SMBHB as a function of GW frequency, with and without BAYESEPHEM (solid blue curve and dashed red curve, respectively). At the lowest frequencies (﹩{f}_{\mathrm{gw}}\lesssim 4\,\mathrm{nHz}﹩), the analysis with BAYESEPHEM was more sensitive than the analysis without it, but there was no difference in sensitivity at higher frequencies.