User research screener

The questions

Each question carries an extraction hint: plain language telling the AI what to pull out of the answer. That's how a spoken ramble becomes a structured row in your inbox.

1. 1Walk us through the last time you dealt with the problem this product is about. When was it, and what exactly happened?

   Voice · 120s

   AI extracts: recency: string, concrete_incident: boolean, incident_summary: string

2. 2How do you handle it today? Name the tools, spreadsheets, or workarounds you actually use.

   Voice · 90s

   AI extracts: current_tools: string[], workaround_described: boolean

3. 3What is the most annoying part of your current way of doing it, and what have you tried to change?

   Voice · 90s

   AI extracts: pain_point: string, attempted_changes: string[]

4. 4How often does this problem come up for you?

   Choice

   Daily / Weekly / Monthly / A few times a year

   AI extracts: selected frequency

5. 5Anything else we should know, plus your general availability for a 30-minute call.

   Text

   AI extracts: availability: string, notes: string

What the AI scores for

This rubric ships with the template and feeds the 1 to 100 score on every submission. It's yours to edit, and the AI writes its reasoning out next to each score.

Score for lived experience, not opinions. High: recounts a specific recent incident in past tense with real detail (names of tools, what broke, what it cost them), and describes a workaround they actually use today. Medium: real experience but stale (a long time ago) or described only in general terms. Low: hypothetical language ("I would", "people probably"), advice-giving instead of storytelling, or answers that suggest they want the incentive rather than have the problem. Flag professional survey-taker patterns.

Try it live

This is a real, working copy of the template. Answer it by voice if you like: submissions here are demo data, processed and then discarded.

Prefer a full tab? Open /f/tmpl-research →