The “right to be forgotten” constitutes a fundamental element in the preservation of privacy. This right empowers users to request the deletion of their personal data from information service providers. While some prior studies have explored the problem of removing a client’s contribution in federated learning, there is a dearth of research considering the unlearning from the user’s perspective. In enterprise-level federated learning, a company participant may possess data entries belonging to millions of users, making deletion requests from users highly concurrent and persistent. Meanwhile, due to the distributed nature of federated learning, removing a user’s data necessitates a collective effort, where uneven deletion requests among different clients and the insecure atmosphere may diminish the motivation for compliance. To address these challenges, we design an efficient and sustainable framework that collaboratively supports user deletion requests within a solid coalition. We conduct unlearning through a closed-form update inspired by the influence function, which is executed by the server using client-provided information thus simplifying the cooperation process. As a result, the unlearning coalition can stably exist throughout the life cycle of federated learning. Experiment results demonstrate the effectiveness of the unlearning algorithm in protecting user privacy.