We should have statistics around plan stage usage for queries overall (e.g. number of INDEX_SCAN, DISTINCT_SCAN, COLLSCAN, etc.) available in server, ideally with additional info like index use, or maybe even alternate plans generated. This would help us in diagnosing issues and tracking statistics about how often various optimizations actually get used in practice, and is especially helpful in cases where it is not obvious (eg via query shape) what plan might result.