One version is approved and another publishes
Sign-off is given by email against a link. Three more changes get made afterwards and nobody asks again. What published is no longer what was reviewed, and there is no way to show what was approved.
Every approval anchored to a specific version, with author, date and timestamp.