Turn exit ticket evidence into Monday's opener

Turn exit ticket evidence into Monday's opener

Most weekend planning with AI works backwards. The teacher asks the model for a Monday lesson, and the model produces something generic because it has nothing to work from. The better Saturday move is to feed it the evidence you already collected last week and let it find the pattern worth re-teaching.

The prompt

You are an experienced [SUBJECT] teacher and assessor in an Australian [PRIMARY / SECONDARY / TERTIARY] classroom. You read student work for misconceptions, not for marks.

I am planning Monday's opener for my Year [LEVEL] [SUBJECT] class. The unit we are in: [ONE SENTENCE describing the unit and where we are up to, e.g. "Year 8 maths, midway through linear equations, students have done one-step and two-step solving and started variables on both sides"].

The curriculum link: [ACARA v9.0 content descriptor, VCE study design dot point, or your own success criteria, copied verbatim].

Below are the responses my [NUMBER] students gave to the exit ticket question:

"[PASTE THE EXACT QUESTION YOU SET]"

Responses (one per line, lightly anonymised):
[PASTE THE RESPONSES VERBATIM. Keep wrong answers. Keep blanks. Keep one-word answers. Do not clean them up.]

Do the following:

1. Cluster the responses into 3 to 5 groups by the underlying thinking, not by whether the answer was right or wrong. Name each cluster in plain English (e.g. "treated the equals sign as 'and then'", not "Cluster 1"). For each cluster, state how many students are in it and quote one short representative response.

2. Identify the ONE cluster worth re-teaching first on Monday. Justify in one sentence: why this misconception, ahead of the others, blocks the next thing in the unit. If two clusters tie, say so and recommend which to address in the opener versus which to address in a small group later in the week.

3. Design a 10-minute Monday opener that directly addresses that cluster. The opener must:
   a. Open with ONE worked example or board move that surfaces the misconception in front of the whole class (not a re-explanation of the correct method).
   b. Include ONE short student-doing activity (under 4 minutes) where students have to commit to an answer (mini-whiteboards, fingers, paired think-aloud, one-line written response).
   c. End with ONE check-for-understanding question that would land DIFFERENTLY for a student still holding the misconception versus a student who has shifted. Write the question. Tell me what the two responses would look like.

4. Then, under "What I am NOT addressing this lesson", list the other clusters and one sentence each on how I should pick them up later in the week (small group, homework re-do, faculty-level conversation).

5. Finally, "Sanity check": one sentence on what would make you, the model, wrong about the cluster you picked. Be specific. If the cluster is really an artefact of the question wording rather than student thinking, say so.

Australian English throughout. Do not use the phrases "deep dive", "unpack", or "scaffold their thinking".

What it does well

Forces the model to cluster by thinking rather than by right-or-wrong, which is the move that turns a pile of exit tickets into a teaching decision. The "Sanity check" line at the end is the bit that stops you walking into Monday convinced of a misconception that is really just a badly worded question.

Where it falls over

The model is reading the responses cold, with no idea which student wrote what. If your strongest student gave a one-word answer because they were rushing for the bell, the model will misread that as a misconception. You still have to look at the names against the clusters before you commit. The prompt also will not pick the right opener for a class you have decided to push forward rather than consolidate. That call is yours.

Try it like this

For Year 7 maths (equivalent fractions exit ticket): paste both the numeric answers and any working students showed. The clusters that surface usually include "added the same number top and bottom", "treated the bar as a subtraction", and "guessed because they were tired". The third one is real, and the prompt should not invent a misconception for it.

For VCE English (a short analytical sentence as an exit ticket): paste the sentences verbatim, typos and all. The clusters tend to split into "summary masquerading as analysis", "evidence without interpretation", and "two arguments in one sentence". The opener writes itself once you can see the split.

If you only have 15 minutes today, do not bother running this on a class you taught well last week. Run it on the lesson that ended with you slightly unsure where the class actually was. That is the one Monday's opener should be built from.

This is the kind of teacher-with-AI move worth showing the head of department on Monday morning. The artefact is the cluster table, not the opener.