Spider millisecond pulsars are, along with some eclipsing post-common envelope systems and cataclysmic variables, part of an expanding category of compact binaries with low-mass companions for which puzzling timing anomalies have been observed. The most prominent type of irregularities seen in them are orbital period variations, a phenomenon which has been proposed to originate from changes in the gravitational quadrupole moment of the companion star. A physically sound modelling of the timing of these systems is key to understanding their structure and evolution. In this paper we argue that a complete timing model must account for relativistic corrections as well as rotationally and tidally induced quadrupole distortions. We solve for the resulting orbital dynamics using perturbation theory and derive the corresponding timing model in the low eccentricity limit. We find that the expected strong quadrupole deformation of the companion star results in an effective minimum orbital eccentricity. It is accompanied by a fast periastron precession which, if not taken into account, averages out any measurement of the said eccentricity. We show that, with our model, detection of both eccentricity and precession is likely to be made in many if not all spider pulsar systems. Combined with optical light curves, this will allow us to measure the apsidal motion constant, connecting the quadrupole deformation to the internal structure, and thus opening a new window into probing the nature of their exotic stellar interiors. Moreover, more accurate timing may eventually lead spider pulsars to be used for high-precision timing experiments such as pulsar timing arrays.