Teachomatic what to automate, and what not to

Lesson Planning: Hand Over, Keep

Lesson planning is two activities wearing one name.

One is producing the artefact: the document, the slides, the worksheet, the objectives in the format your school wants. The other is working out what the lesson is — deciding what the difficulty actually is for this class, choosing the example that will land, anticipating where they will go wrong.

For a workplace-oriented comparison outside education, the background offers another way to look at measurement, workload, behaviour, or accountability.

The first is admin. The second is the job. And the reason this matters more than the usual "keep the human in the loop" advice is that the second one is invisible in the output. A plan produced without it looks identical to a plan produced with it, right up until the lesson.

What hands over cleanly

The format. Turning a plan you have in your head into your school's template. Objectives phrased in the required language, timings, differentiation column, the boxes that exist for the folder rather than for the lesson.

For an external perspective on lesson planning, teacher workload, and classroom technology, see Canva for Education.

Variations on something that works. You have a good task and need it for a different text, a different period, a different data set. The thinking transfers; only the surface changes.

The material at three levels. Same content, adjusted reading level and scaffolding. This is one of the highest-yield uses there is, and it is mechanical work you would otherwise skip for lack of time.

Alternative explanations. You have used your two analogies and neither worked. Asking for five more is a genuinely good use, because you are the one who will judge which is any good.

Starters, exit tickets, extra practice questions. Low-stakes, high-volume, easy to check — though the sheet that is doing the teaching needs more from you.

What does not

Deciding what the difficulty is. The core planning question — what specifically is hard about this for these students — cannot be answered without knowing the students. A generated plan answers a different question: what is generally hard about this topic. Those diverge constantly, and the divergence is where your lesson either works or does not.

Choosing the example. Examples are where teaching happens. A good one connects to something the class already has, which is knowledge that exists only in your head, built from months of watching them.

Sequencing for this class. The general sequence is in the textbook. The one you need depends on what stuck last week, which is something you learn from reading their work rather than from a summary of it.

Predicting the misconception. You know that a third of them will do the specific wrong thing, because they did it last year. Generic misconception lists are real and shallow; yours is specific and predictive.

The order that protects the thinking

The sequence matters more than the rule, because the danger is not delegation. It is delegation before you have thought.

Think first, without the tool. Five minutes, on paper: what is this lesson for, what is hard about it for this class, what will I do when it goes wrong. Five minutes is enough, and it has to come before you open anything.

Then generate the artefact from that thinking, having told it what you decided — a reusable context block removes most of the retyping.

Then edit for this class, which is where your five minutes gets spent.

Do it in the other order and something specific happens: you get a plausible plan, it becomes the anchor, and you end up editing somebody else's idea of the lesson instead of building your own. It will be an adequate lesson. It will not be your best one, and you will not be able to say what went missing.

The test for a generated plan

Before you use it, three questions.

Where does this go wrong? If the plan contains no anticipation of failure, it was written for an imaginary compliant class.

Which part is doing the teaching? In a good plan you can point at the moment where the understanding is supposed to happen. Generated plans frequently have activities and no such moment.

Would I know it was mine? Not as vanity. If a plan is indistinguishable from any other teacher's plan for that topic, it has not been adapted to your class, and your class is the only reason you are writing it.

The thing nobody warns about

Planning is when you rehearse. Working through the lesson in your head is how you find out you do not actually understand the topic well enough to explain it, or that the transition between parts two and three does not work.

Reading an already-written plan does not do this. It feels like preparation and it produces the document, but you arrive at the lesson never having walked through it.

The symptom is specific: the lesson runs fine until a student asks something slightly off-script, and you find yourself with nothing behind the plan. That gap is the rehearsal you did not do.

Simple countermeasure, whatever else you automate: say the explanation out loud once, to nobody, before the lesson. Two minutes. It catches most of it.

For heads of department

Two things worth setting deliberately.

Shared banks of generated material are a good use. Levelled texts, question sets, model answers — produced once, checked once, used by everybody. That is the efficiency case and it is real.

Requiring plans in a format nobody reads is now actively harmful. It always wasted time; now it is the part that automates perfectly, which means the folder fills with documents nobody wrote and nobody reads. If the plan is a compliance artefact, say so and stop pretending it is planning.

The short version