It’s still, of course, personal data when processing it. When you run a statistics function, its input is personal data, but its output is no longer. The person who can run this is still only the data operator. They can’t just randomly assign somebody else to run this data processing for it.