Removing duplicates and keeping one row

DISTINCT cannot deduplicate customers: the two registrations of one person differ in customer_id, signup_date and the spelling of the email, so they are distinct rows. What you need is a survivor rule: within each group of rows that describe the same thing, which one do you keep?

Common rules are the earliest record (it owns the longest history), the most recently updated record (it has the freshest contact details) or the most complete record. Whatever you choose, write it down and make it deterministic: if two rows tie on the rule, a tie-breaker (usually the id) decides, so the query gives the same answer every time.