Research Guard questionnaire-review POC — run 1

Raw source preview

Raw, unprocessed file text shown below.

---
title: "Research Guard questionnaire-review POC — run 1"
status: processed
source_type: prototype test
session_date: 2026-03-31
duration: "47m 09s"
participants:
  - Danny Bowles
  - Desmond Leo
  - Gary Walsh
  - Carlos Montoliu Ducay
product_area: "Research Guard"
tags: [research-guard, questionnaire-review, prototype-test, internal-practitioner]
---

# Research Guard questionnaire-review POC — run 1

## Source file

- [questionnaire-reviewer-poc-run-1.docx](./questionnaire-reviewer-poc-run-1.docx) — original Microsoft Teams transcript export.

## Evidence relevant to launch

- The prototype caught a single-code/select-all conflict; Danny reacted, “Oh my god, good call” (5:55).
- He repeatedly recognised useful review questions and said a second confirmation could be a good sense check for research-team logic, provided the impact and trade-offs were explained (11:22, 18:28, 22:30 and 24:00–25:00).
- He also found parts frustrating, unclear and too confirmation-heavy, particularly for a complex survey (18:28–31:29).
- He wanted the system to work at the strategic-design level, including stakeholder and commercial context, rather than only perform technical checks (41:10–44:32).

## What this supports

The prototype found a consequential issue in a real questionnaire and elicited detailed expert feedback that shaped the next flow.

## What this does not prove

This was an assisted internal POC. It does not prove recommendation accuracy, senior-equivalent output, time saved, production usability or willingness to pay.