The alignment of intrinsically anisotropic olivine crystals under convection is typically invoked as the cause of the bulk of seismic anisotropy inferred from shear-wave splitting (SWS). This provides a means of constraining the interplay between continental dynamics and the deep mantle, in particular for densely instrumented regions such as North America after USArray. There, a comparison of “fast orientations” from SWS with absolute plate motions (APM) suggests that anisotropy is mainly controlled by plate motions. However, large regional misfits and the limited realism of the APM model motivate us to further explore SWS based anisotropy. If SWS is estimated from olivine alignment in mantle circulation instead, plate-driven flow alone produces anisotropy that has large misfits with SWS. The addition of large-scale mantle density anomalies and lateral viscosity variations significantly improves models. Although a strong continental craton is essential, varying its geometry does, however, not improve the plate-scale misfit. Moreover, models based on higher resolution tomography degrade the fit, indicating issues with the flow model assumptions and/or a missing contributions to anisotropy. We thus compute a “lithospheric complement” to achieve a best-fit, joint representation of asthenospheric and frozen-in lithospheric anisotropy. The complement shows coherent structure and regional correlation with independently imaged crustal and upper mantle anisotropy. Dense SWS measurements therefore provide information on depth-dependent anisotropy with implications for tectonics, but much remains to be understood about continental anisotropy and its origin.