The article critiques the concept of a single source of truth in data mesh architecture, arguing that it's not only achievable but essential for business value. The author, a data mesh skeptic, believes that while data mesh principles can enhance centralized architecture, they often lead to confusion during implementation. Many teams fail to coordinate, leading to discrepancies in crucial metrics like revenue. This lack of alignment among domains can result in conflicting definitions and duplicated efforts. The hybrid mesh architecture is proposed as a solution, combining the benefits of decentralized data development with essential centralized governance. In this model, domain teams can create their own data products while relying on a common infrastructure set up by a central data team. The hub-and-spoke pattern allows for shared data models and integration, but practical issues arise when teams face dependencies and overlaps in their data needs. This often leads to complications in orchestration and ownership disputes. To address these challenges, the author suggests a practical approach: centralize shared data. By moving shared information upstream to a central location, teams can establish a single source of truth and avoid the pitfalls of decentralized development. The article emphasizes the importance of clear data ownership and the need to reconcile dependencies among domains, ultimately advocating for a more centralized approach to data architecture.