A major obstacle for computing optical spectra of solids is the lack of reliable approximations for capturing excitonic effects within time-dependent density-functional theory. We show that the accurate prediction of strongly bound electron-hole pairs within this framework using simple approximations is still a challenge and that available promising results have to be revisited. Deriving a set of analytical formulae we analyze and explain the difficulties. We deduce an alternative approximation from an iterative scheme guided by previously available knowledge, significantly improving the description of exciton binding energies. Finally, we show how one can "read" exciton binding energies from spectra determined in the random phase approximation, without any further calculation.The response of materials to an electromagnetic field is a key to many properties and applications. In the frequency range from infrared to ultraviolet, the optical properties determine the color of materials, their ability to absorb the sunlight, and much more. They lay the ground for non-destructive spectroscopies such as ellipsometry, that can tell us much about the electronic or atomic structure of materials. However, theoretical tools are needed that allow one to analyze, understand and predict measured results and desired or undesired properties. These tools should be reliable and versatile, but simple enough to be applicable to systems of fundamental or technological interest, that are often rather complex. One of the major challenges is to design approximations for the ab initio calculation of optical spectra of extended systems such as solids and liquids [1].The state-of-the-art approach for the ab initio calculation of optical spectra consists in using the Kohn-Sham (KS) electronic structure coming from a density functional theory (DFT) calculation as starting point for a quasiparticle bandstructure calculation in the GW approximation, and the subsequent solution of the BetheSalpeter equation (BSE) to account for the electron-hole interaction [1][2][3][4][5]. The scheme is successful; in particular, excitonic effects are well described. However, calculations are computationally demanding, because of the two-particle (electron and hole) nature of the problem. Alternatively, time-dependent DFT (TDDFT) [1,6,7] formulates the response in terms of variations of local potentials that are functionals of the time-dependent density. This reduces the size of the problem, but raises the question of how to find a good approximation for the time-dependent exchange-correlation (xc) potential v xc and its first derivative, the xc kernel f xc (r, r ′ , t − t ′ ) = δv xc (r, t)/δn(r ′ , t ′ ), where n is the time-dependent electron density. Some simple but widely used approximations such as the adiabatic local density approximation [8,9], that are often successful for finite systems and for electron energy-loss spectra, yield disappointing results similar to the random phase approximation (f xc = 0) [7,10] for absorption spectra of solids.Many wo...