Practice: Repair a Weak Output Through Three Follow-ups

Unit ID: PAW-M07-U05 Estimated active time: 25-30 minutes

What you are producing

One weak output, three targeted follow-ups, and a record of what each one changed.

Start from something genuinely weak

Use a real disappointing output if you have one. If not, make a deliberately thin request — a one-line ask for something that needed context.

Do not fix it in the first prompt. The exercise is the repair.

Fix one thing per follow-up

Use the triage from unit 07.00. Deal with facts first, then shape, then judgement — in that order, because fixing shape before facts wastes the effort.

The repair record

REPAIR RECORD

Original request:
Output problems (triaged):
   facts:
   shape:
   judgement:

Follow-up 1 (facts):        what I asked        what changed
Follow-up 2 (shape):        what I asked        what changed
Follow-up 3 (judgement):    what I asked        what changed

Total time:
Would a better first prompt have been faster?   yes / no
Lines to promote into the saved prompt:

Answer the honest question

The "would a better first prompt have been faster" line is the one that matters.

Often the answer is yes, and the three follow-ups have just shown you exactly what the first prompt was missing. That is the finding.

What to keep

Save the repair record and add its promoted lines to your improvement log.

This is the third portfolio artefact. The capstone in Module 12 asks for one of these against a real task.

Mark complete