Correction - statistically they're not in the audience. i.e. even if you'd one person would be a good Candy Japan customer, YouTube ad billing incentivizes you to lower the false-positive rate (ie the number of people who view your ad but don't buy) rather than the false-negative rate (ie The number of people that skip your ad but would have been interested in the product).

Sometimes you get it wrong, like with GP, but that's worth it.

I think this is a Good Thing - it makes advertisers and YouTube care in dollars-and-cents terms about not wasting ad viewers' time and attention.