This research compares the relative effectiveness and efficiency of personalised feedback versus annotated exemplars as two approaches to providing high school students with formative feedback in the development of their writing skills. The study was a randomised experiment conducted in situ in two English classes in one high school in New Zealand. The primary researcher was the teacher in the study. The findings show that there was substantial growth in writing performance with both approaches. Students appreciated both approaches, but preferred personalised feedback. Depending on what was included in the calculations, providing personalised feedback took between two and eight times as much teacher time to deliver than annotated exemplars. The theoretical and practical implications are discussed.