E-commerce index strategy

Design the indexing strategy for an e-commerce orders table (2 billion rows, 25k writes/sec) that must serve several query shapes fast: filter by customer_id + date range, filter by status + warehouse, full-text search on product name, and a dashboard counting orders per status per hour. Adding indexes blindly is killing write throughput. How do you choose which indexes to build, order their columns, and avoid index bloat?