Mom Test

Mom Test Questions

Generate questions that follow Rob Fitzpatrick's The Mom Test rules: ask about the person's life and past behavior, not your idea. Good questions can't be lied to.

The Three Rules

1. Talk about their life, not your idea
2. Ask about specifics in the past, not hypotheticals about the future
3. Talk less, listen more

Inputs

When invoked, you need:

• The idea/feature/request: a GitHub issue, Linear ticket, Slack message, or plain description (REQUIRED)
• Your hidden question: what the user actually wants to validate — ask if not obvious from context (OPTIONAL but valuable)

Question Categories

Generate 2-3 questions per category. Not all categories apply to every situation — skip irrelevant ones.

Current workflow

How they do things today. Reveals whether the problem is real and frequent.

• "Walk me through the last time you [did the thing the feature addresses]. What happened?"
• "How often does [situation] come up?"
• "What did you do when [the gap] happened?"

Pattern: "Walk me through..." and "The last time you..." force specifics.

Pain

How much the current situation actually costs them. Reveals severity.

• "How much time did you spend [workaround] last time?"
• "Have you tried any workarounds?"
• "What happened when [downstream person] couldn't [thing the feature would enable]?"

Pattern: Ask about consequences they've already experienced, not ones they imagine.

Stakes

Whether this matters enough to change behavior. Separates "nice to have" from "need to have."

• "If [tool] never supports [feature], what would you change about your workflow?"
• "How many of your recent [sessions/projects/sprints] hit this?"
• "Who else on your team runs into this?"

Pattern: "If we never built this..." reveals true priority. Low-stakes answers: "I'd just keep doing what I do."

Solution assumptions

Whether the requester's proposed solution matches the actual need. People jump to solutions — dig for the underlying problem.

• "When you say [their proposed solution], would you still want to [deeper interaction], or is [simpler thing] enough?"
• "What kind of [input] are we talking about — [simple case], [complex case], or [edge case]?"

Pattern: Offer a deliberately reduced version. If they accept it, the real need is smaller than the request.

Alternatives already rejected

What they've tried and abandoned. Reveals hidden constraints.

• "Have you looked at other tools for [this]? What was wrong with them?"
• "You mentioned [alternative approach] — what made you move away from that?"

Pattern: Past tool-switching decisions reveal real priorities better than stated preferences.

Sharing vs. personal

Whether the artifact is for them alone or shared with others. Changes the feature's scope entirely.

• "Who else sees [the artifact] besides you?"
• "Walk me through what happens to [the artifact] after you're done with it — does it go anywhere?"
• "If I looked at your last [project] that had [the artifact], would [other person] have opened it?"

Pattern: Ask where the artifact goes, not whether they share it. The third form grounds it in a specific past instance.

Writing Good Questions

| Do | Don't | |---|---| | Ask about past behavior | Ask about future intentions | | Use "the last time you..." | Use "would you ever..." | | Ask who else is affected | Ask if others might be interested | | Offer a reduced version | Ask if they want the full version | | Ask what they did when X happened | Ask what they'd do if X happened | | Ground in a specific instance | Accept vague generalities |

Output Format

Group questions by category with a brief note on what each group validates. End with a one-line note identifying the single most important thing to validate (the "if this is false, nothing else matters" question).

When This Skill Is Wrong

• You have quantitative data (analytics, usage logs) — look at that first
• The requester is reporting a bug, not requesting a feature — just fix it
• You're interviewing yourself — Mom Test is for external validation

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: tomasz-tomczyk
• Source: tomasz-tomczyk/skills
• License: MIT

Install and usage instructions live in the source repository linked above.