What are the "success rates" of competing approaches in which there is a large sample? Without that, it is hard to make much sense of your eight to twelve percent.