Executable User Research
Use this when you’ve observed a problem and want working alternatives in front of real users within days. Worked in “AI Helps Most When It Removes Handoffs Between Users and Builders.”
Executable User Research: one line per step, with what happened and the evidence it produced.
| Step | What happened, and who did it | Evidence it produced (User Evidence Ladder level and UAT Distance) |
|---|---|---|
Observe real work |
||
State the hypothesis |
||
Build two or three prototypes against realistic data |
||
Put them in front of real users at D0–D1 |
||
Measure |
||
Decide |
The third column keeps the loop honest: until the fourth row names real users at D0–D1, nothing in it rises above the team’s own opinion. The Alternative Prototypes Prompt and the AI-Assisted UAT Design Template in Appendix B help with the third and fourth rows; nothing helps with the first except going to watch.