<p>Thermal phase variations of exoplanets are a patent testimony of their multidimensional nature: day-to-night temperature contrasts range from hundreds to thousands of degrees. &#160;Nonetheless, the spectra of these planets have typically been fit using 1D retrieval codes that only account for vertical temperature gradients. &#160;Recent multi-dimensional retrieval schemes are generally based on linear combinations of 1D models, which are more liable to degeneracies and more computationally demanding. &#160;Here we present an alternative: phase-dependent spectral observations are inverted to produce longitudinally resolved spectra that can then be fitted using standard 1D spectral retrieval codes. We test this scheme on the phase-resolved spectra of WASP-43b and on simulated JWST observations using the open-source Pyrat Bay retrieval framework. &#160;We show that 1D spectral retrievals on longitudinally resolved spectra are more accurate than applying 1D spectral retrieval codes to disk-integrated emission spectra, highlighting the impact of longitudinal variations in composition in addition to temperature. &#160;In particular, we find that JWST phase measurements of WASP-43b should be treated with longitudinally resolved spectral retrieval.</p>