Confusing language (to me)
She continues, “We also purchased third-party U.S. demographics data last month that you can use to fill in missing values. In fact, I don’t want you to leave any missing data, outliers, or bad data blank in the customer data. Rather, replace missing data, outliers, and bad data with estimated values from the U.S. demographics dataset based on the customers’ zip codes.
“However, you must delete missing data, outliers, and bad data in the email campaign and social media dataset because we have no way of looking up that information in another dataset.