"Anonymization". Two reasons: either they de-anonymize the data, which is a good exercise for them but still approximate, or the demographic data is merely registered beforehand. Here, I think they for instance need the age (and not its category) because of discounts to young and old people.