A polished lesson can conceal this mismatch until the final quiz. Reviewing the requested evidence early gives you something specific to change.
Name the performance you need to see
Consider an original fictional workshop about evaluating claims. Its objective is: "Given two claims and their supplied evidence, choose the better-supported claim and justify the choice with one relevant observation and one limitation."
The proposed final question is "Define evidence." A correct definition does not show whether the learner can compare claims, select relevant information or recognize a limitation.
Carnegie Mellon University's assessment-alignment guidance, checked October 24, 2026, connects objectives, assessments and instructional activities. Use that relationship to identify the mismatch; it does not certify this fictional workshop or prove its effectiveness.
Write a task that elicits the missing evidence
Use this small authored case. Both claims concern a supplied log of 12 deliveries from one route last Tuesday:
| Claim | Supplied evidence | What the evidence supports |
|---|---|---|
| A: All our routes are always on time. | The log marks 10 of the 12 deliveries on time | Does not support the universal claim |
| B: Most deliveries in this route's Tuesday log were on time. | The same log marks 10 of 12 on time | Supports this bounded description |
Ask learners to choose A or B and explain why. A model response for the stated objective is: "B is better supported because 10 of the 12 logged deliveries were on time. The log covers only one route on one Tuesday, so it does not establish the pattern across all routes or dates."
That response contains a choice, a relevant observation and a limit. It does not establish why two deliveries were late, whether the record is accurate or how next Tuesday will go. Preserve those boundaries in the answer key.
Ask for an alignment audit, then a revision
Audit this instructional design before drafting teaching materials.
Learners and prerequisites: [ENTER]
Observable objective: [PASTE]
Proposed final assessment and scoring criteria: [PASTE]
Planned practice: [PASTE]
Permitted case evidence and its limitations: [PASTE]
Required school or organization format: [ENTER]
For each part of the objective, identify the learner response that
would demonstrate it. State what the assessment currently elicits
and what it leaves untested. Propose a revised task and criteria.
Then check whether the planned practice prepares learners for that
task. Do not invent outcomes, standards alignment or learner scores.
Preserve the subject facts and flag decisions for the instructor.
Return the audit before drafting slides or a timed lesson plan.
OpenAI's prompting documentation supports supplying source material and specifying the result. Here, the useful result is a trace from each objective requirement to observable learner work.
Try the task without its model answer
Have a subject reviewer answer the revised question from the case alone. Check whether more than one response could satisfy the criteria and whether the key accepts an accurately stated limitation. For this exercise, quoting "10 of 12" without bounding the claim leaves one part of the objective incomplete.
Practice should also involve choosing and explaining, with feedback. Repeatedly reading definitions would not rehearse the full assessment task. Use another permitted case during practice so the final response does not merely reproduce the model wording.
Once the evidence demand is settled, lesson planning under a time constraint addresses the next design problem. If you already have a reviewed lesson, checking companion materials helps keep its handout and key consistent. Approve the revised assessment before investing in the lesson's production format.