Note how it says "This estimate includes episodes with a stranger or slight acquaintance perpetrator in which a child was moved at least 20 feet or held for at least 1 hour"
Obviously a case where a child was moved only 20 feet, or held for less than a day belong in the 99.8% of cases where the child is returned. Even if the child is held for 1 month, that's going to be part of the 99.8% of cases where the child is returned. And, a kid would have to be moved much more than 20 feet to be missed by a search party. So, presumably the majority of "stereotypical kidnappings" result in the child being returned.
What it has to do with what we were originally talking about is that the 0.2% that aren't returned aren't 100% "stereotypical kidnappings". The vast majority of that tiny minority are also a kid running away, or a kid with a non-custodial parent. Only a tiny fraction of that tiny fraction (so maybe in the single digits per year) are kids who are kidnapped by a stranger or slight acquaintance and who never return.