I study demand for social media services by conducting an experiment where I monitor how participants spend time on digital services and shut off access to Instagram or YouTube on their mobile phones. I characterize how participants substitute their time during and after the restrictions, which motivates estimating a model of time usage with inertia. I apply the substitution patterns observed during the restriction period and implied by the model with and without inertia to a relevant market definition exercise to conclude that relevant markets may be larger than those considered by regulatory authorities for social media applications.