I’m not sure if it is confusing. I guess one way would be to actually test on someone.
Give helper labels that assist in understanding without changing the definition of an npub;
And as a secondary experiment remove the signup step entirely and just generate keys for user, then explain what just happened.