This is the most common antipattern when SLPs use LLMs for goal writing. The output reads well (complete sentences, professional vocabulary, appropriate-sounding structure) but it fails the fundamental question: How would you take data on this?

The Bad Example

“Student will demonstrate improved pragmatic language skills by engaging in more appropriate social interactions with peers across various school settings as observed by the speech-language pathologist.”

This goal sounds great. A parent reading it would think, “That seems thorough.” An administrator might approve it without question.

But try to take data on it. What counts as “improved”? What’s “more appropriate”? What does “various school settings” mean: two? Five? The cafeteria and the playground? “As observed by the SLP”… during what? A 30-minute session? A full school day?

Why LLMs Produce This

Language models optimize for fluent, professional-sounding text. Measurability is not a property the model evaluates. It has no concept of “can a clinician actually track this?” It generates language that looks like a good goal because it matches the surface pattern of goals in its training data.

The Fix

After any LLM generates a goal, apply the Three-Question Test:

1. What exact behavior am I counting? (If you can’t name it, it’s not measurable.)
2. What does success look like as a number? (Percentage, frequency, duration, or count.)
3. Could another SLP take the same data and get the same result? (If not, the criteria are too subjective.)

The Fixed Version

“Given a structured small-group activity with 2-3 peers, [Student] will initiate a topic-relevant conversational turn (comment, question, or response) in 4 out of 5 opportunities across 3 consecutive sessions, as measured by SLP data collection during weekly pragmatic language group.”

Now you know: what (topic-relevant conversational turn), how much (4/5 opportunities), how consistently (3 consecutive sessions), in what context (small-group, 2-3 peers), and who measures (SLP during group).

The Takeaway

If a goal reads beautifully but you can’t picture yourself taking data on it during a session, the LLM did its job (writing) and you need to do yours (clinical measurement design).

Pair With

• IEP Goal Strengthener prompt – the prompt that fixes this
• Writing Goals tasks – the full goal-writing framework
• Schools setting guide – why measurability is legally required under IDEA