Data is not localized to just a single product team. If you think about it, every product is designed to produce data. The challenge that so many organizations have is realizing that data access and usage are already spread across every organization and skill level at their company. Thus, it seems that every agile scrum team has a need for thought leadership and data strategy best practices. Rather than turn those data needs over to a single downstream team, executives embedded data scientists within each software team. This kept the ownership of the solution on the team that produced the data. It also allowed data scientists the flexibility to research cross-product problems (like machine learning) in a centralized team.
This team is a lightweight product discipline. Each data scientist is responsible for the utilization of data within their product domain. A data scientist can function on multiple product teams, and all data science work is prioritized across the entire new product development strategy. This has ensured our company is focusing on the most important data initiatives, simultaneously keeping product managers focused on other deliverables.