Hi again,
Mid-September. Someone in the workroom says "progress reports are in a few weeks" and keeps walking, like that's a neutral piece of information.
Your stomach drops anyway. Because the goal says "measured biweekly," and the data binder you set up in August — the one with the labeled dividers and the clean graphs — is still mostly aspirational. It's week five. The graph has one dot.
Nobody trained us for this part. We got trained to write goals. Collecting the evidence that a student is actually moving toward one, in the middle of teaching, every week or two? You were supposed to just... figure that out.
The goal already did the hard part
Here's what I wish someone had told me years ago: a well-written goal already contains its own measurement system. The condition, the behavior, the criterion, the schedule — it's all sitting in that one sentence. You don't have to invent a system. You have to unpack the one the goal already defines.
And unpacking is exactly the kind of job AI does well. One planning period, one prompt, and you walk out with the whole machine: what to measure, how often, a tracking sheet you can rebuild in five minutes, a routine that doesn't stop instruction, and the checkpoints where a flat line should get your attention.
First, make the goal anonymous
A goal as written has a name in it. That name never goes into the AI. You know the rule by now, and it doesn't bend.
The fix takes ten seconds, because everything the AI needs is in the rest of the sentence. "Given a 5th-grade-level passage, [Name] will read 110 words per minute with 95% accuracy in 3 of 4 trials" becomes "Given a 5th-grade-level passage, the student will read 110 words per minute with 95% accuracy in 3 of 4 trials."
Swap the name for "the student." Drop anything else that points at a real kid. What's left is a sentence about a skill and a standard, and that's all the system needs to be built from.
Then paste this:
You are an experienced special education teacher setting up progress monitoring. Here is an IEP goal with every identifying detail removed: [paste the de-identified goal, e.g., "given a 5th-grade-level passage, the student will read 110 words per minute with 95% accuracy in 3 of 4 trials"]. Design a simple system: (1) exactly what to measure and how often, matched to the goal's own criteria; (2) a paper or spreadsheet tracking layout I can recreate in five minutes; (3) a realistic in-class routine for collecting the data without stopping instruction; (4) the checkpoints where I should worry if the trend is flat. Don't change the goal, and don't invent any student details.
The response will be long. The tracking sheet is the part you rebuild today; the rest is reference.
What comes back (and what you still own)
For that fluency goal, mine came back with a one-minute timed read twice a week, a five-column tracking sheet (date, passage level, words per minute, accuracy, trial count), a routine that tucks the timed read into the last two minutes of small-group time, and a note to look hard at the trend around week four.
Read it like the professional you are. Check the schedule against what the goal and your district's paperwork actually require. Sanity-check the routine against your real classroom — the AI has never met your third period. And if your district has an AI policy, that's the final word, same as always.
One boundary worth naming: if the data does come back flat, what happens next — adjusting supports, services, the goal itself — is an IEP-team conversation, not a prompt. The AI builds the tracker. The humans read it and decide.
You don't need to invent a data-collection system. Every measurable goal already is one — folded up. You're just unpacking it.
The part nobody tells you
The empty binder's real weight was the low-grade guilt every time you walked past it. Week five of "I'll start next week" weighs more than the data collection itself ever would.
The first data point ends that. Not because one dot proves anything — it doesn't — but because a system that has started is a completely different thing to live with than a system you keep meaning to start. You stop avoiding the binder. And data you collect in September is worth ten times what you try to reconstruct in October, the week reports are due.
Try it this week
Pick one goal. Just one. The one whose empty graph bugs you most. De-identify it, run the prompt during a planning period, rebuild the tracking sheet in your own document, and collect one data point before Friday. That's the whole assignment.
Then hit reply and tell me how you actually collect data right now — clipboard, sticky notes, an app you half-use, straight-up memory? No judgment; I've done all four. Your answers shape what I write next.
Talk soon,
Shaun
P.S. Forward this to the teacher down the hall whose binder looks like yours. September is easier with company.
A quick but important note: this newsletter is for general educational and informational purposes only. I'm a special education teacher sharing time-saving ideas. The views here are my own and don't represent any school district or employer. I'm not a lawyer, and nothing here is legal, compliance, or professional advice. Always follow your district's and state's policies, especially on AI use and student privacy, and check with the right person in your district before acting on anything significant. You are the professional: review and finalize everything yourself. And never put student names or identifying details into any AI tool.