[skeeter2020]

I've always found that this is a great idea that doesn't hold up to the reality of DB tuning. There's no "typical" query, in the sense that should it optimize for the more common but pretty performant query, or the rarer, very slow query? What type of client are you most concerned about? Both? Is that possible? which queries are worth the (total) cost of the index? Does a specific index hurt some other workload like automated integrations, or is it OK to have slow updates in the middle of the night? You can definitely profile over time and there are tools that help you capture data and make recommendations, but auto-indexing would be a limited to poor solution a lot of the time.