A unit plan, not just a lesson

Lifting the view

Everything so far has been one lesson at a time. A term is not a sequence of lessons; it is a structure, with prerequisites that have to come first, ideas that need to be returned to, and moments where you find out whether the last three weeks landed before you build on them. Planning at that level is what experienced teachers do in their heads and new teachers are rarely taught.

The model can draft that structure, and its draft has one predictable flaw.

The prompt

Plan a 12-week unit for Class 9 geography on rivers and flooding, 3 lessons a week of 40 minutes, following this syllabus extract: [paste].
Show, for each week: the topic, what it depends on from earlier weeks, one earlier topic revisited in the starter, and how I will check understanding that week (a short written check, not a discussion).
Put a low-stakes test in week 4 and week 8, each covering everything so far.
Mark any topic that students in your experience find hard, and say why.
Plain text.

The middle three lines are the ones the model would not include by itself.

The flaw in the default

Ask for a unit plan without those lines and you get the check question: a new topic every week, in textbook order, a test at the end. It is the average of every scheme of work on the internet, and most of those are linear because they were written to cover content, not to build memory.

A plan built for learning looks different. Topics return. Week 5’s starter retrieves week 2. The test in week 4 exists so that you find out about a misconception before week 6 is built on top of it. The model produces this structure readily when asked and never when not asked, which is the same lesson as every other in this course: it does what the average does unless told otherwise.

The two things it cannot know

Your calendar. Festivals, public holidays, exam weeks in other subjects that take the hall, the sports day, the week of the inspection, the monsoon week when half the class is absent. The model plans twelve uninterrupted weeks. Your term has nine. Paste the school calendar and say “weeks 3 and 7 are lost; replan.” Or accept the twelve-week draft and cut it yourself, which takes ten minutes and is the ten minutes where the planning judgement lives.

What your students have found hard. The model’s “students find this hard” is a guess from general teaching writing, and it is often right. Your record from last year is better. The topic where the Class 9 of two years ago fell apart deserves two weeks and an early check, and only you know which topic that was. Put it in the prompt: “Last year students struggled with hydrographs; give it extra time and an early check.”

Prerequisites and the missing lesson

One useful thing to ask: “For each week, list what a student must already know, and flag anything on that list that is not covered earlier in this plan or in the previous year’s syllabus.” This surfaces the missing lesson: the prerequisite everybody assumes and nobody teaches. Teachers who run this on a unit they have taught for years often find one.

From unit to lessons

Once the unit plan is right, each week’s line is the “what came before” and “what comes next” for that week’s lesson prompts. Paste the week’s line at the top of each lesson request and the lessons stay in sequence. The unit plan becomes the context that keeps every generated lesson pointing the same way, which is the fix for the sameness and drift the earlier lessons described.

Sharing a scheme

A department scheme of work built this way (with revisits, checks, and the “students find this hard” notes) is a document worth keeping and improving each year. Add a column for what actually happened: the week that ran long, the check that revealed a problem, the topic that needed a fourth explanation. The notes are what make next year’s plan better than this year’s, and they are the part the model cannot write.

The honest limit

A unit plan is a hypothesis. The class in front of you will be faster in some weeks and slower in others, and a plan followed rigidly because it is written down is worse than no plan. Build the revisit-and-check structure, then treat the sequence as something you will adjust every Friday. The model gave you the skeleton in a minute; the adjusting is the job, and it is yours.