You are given raw customer feedback survey data from American Family Insurance touchpoints, along with policyholder and channel reference tables. The dataset contains inconsistent casing, blank strings, duplicate submissions, invalid score values, and incomplete responses. Write a PostgreSQL query that returns one cleaned record per customer for the most recent valid survey submission, standardizes key fields, flags whether the response is complete, and shows the previous valid score for that customer when one exists.
Treat blank strings as missing, normalize response_channel to the mapped channel name when possible, convert score values outside the 0-10 range to NULL, and exclude test submissions. If a customer has multiple submissions on the same day, keep the latest by timestamp and then highest response_id as a tie-breaker.
