California researchers in the early nineteen-eighties conducted a five-year study of teacher-skill development in eighty schools, and noticed something interesting. Workshops led teachers to use new skills in the classroom only ten per cent of the time. Even when a practice session with demonstrations and personal feedback was added, fewer than twenty per cent made the change. But when coaching was introduced—when a colleague watched them try the new skills in their own classroom and provided suggestions—adoption rates passed ninety per cent. A spate of small randomized trials confirmed the effect. Coached teachers were more effective, and their students did better on tests.