The Mom Test: A User Interview Method for Validating Demand

The JTBD statement from the previous chapter is still a hypothesis. Interviews help us check whether the problem has occurred recently and whether users have already spent time, money, or effort on it.

This sounds straightforward. Yet if we directly ask whether someone likes an idea, we often receive polite, general replies. For example:

• What do you think of this idea?
• Would you use this if I built it?
• Does this feature sound useful?

The replies usually sound encouraging:

• Sounds good
• That seems useful
• I think you should try it

The problem is that these answers usually do not help you decide anything. They are often just politeness, support, or a natural instinct not to discourage you in the moment. You think you collected "market validation," but what you really collected was a pile of comforting feedback that is hard to use.

These answers are not wrong, but they are difficult to use for product decisions. The Mom Test asks us to spend less time on future intentions and more on what happened in the past, how the user handled it, and what it already cost them.

This chapter explains how to avoid such false-positive feedback and shift the interview from opinions and hypothetical futures toward past behavior, current workarounds, and costs already incurred.

1. Why Interviews Produce False Positives

The Mom Test comes from Rob Fitzpatrick's book of the same name. The title sounds playful, but the point is sharp:

even your mom will struggle to tell you your idea is bad if you ask the wrong way.

The reason is not that she is dishonest. It is that:

• she does not want to hurt you
• she naturally wants to encourage you
• she will often answer in the direction your question already suggests

And this is not only about your mom. Friends, coworkers, former classmates, and even strangers often do the same thing when they react to a product idea. A positive answer does not necessarily mean the demand is real. It may simply mean you asked in a way that made a flattering answer easy.

The point of The Mom Test is therefore not whom to interview, but how to structure the questions. An interview should look for verifiable experiences instead of collecting evaluations of the product idea.

2. Opinions and Behavioral Evidence

The Mom Test mainly helps you avoid one very common cognitive mistake:

mistaking polite positive feedback for real demand.

For example, people often ask:

• What do you think of this app idea?
• If I built an AI tool that rewrites resumes, would you use it?
• Does this feature sound valuable?

These questions have three things in common:

• they ask for opinions
• they contain some amount of suggestion or framing
• they talk about a future that has not happened yet

People are usually unreliable when answering about opinion and imagined future behavior. They tend to overestimate their own interest, their own follow-through, and their own willingness to pay.

That is why The Mom Test keeps reminding you:

• do not trust praise for your idea too quickly
• do not trust predictions about future behavior too quickly
• bring the conversation back to what the user has already done in real life

Compared with "Would you use this?", a question like "How did you handle this last time?" is usually much closer to truth.

3. Three Principles for Asking Questions

If you want to remember only the most important part first, remember these three principles.

3.1 Talk less about your idea and more about the user's real past experience

Many weak interviews start with too much explanation: your solution, your excitement, your product concept, your feature plan. Once you do that, the other person often shifts into "supportive mode."

A better direction is to center the conversation on their real experience:

• When was the last time this happened?
• What were you doing at the time?
• How did you handle it?
• Which step felt the most annoying?

Questions like these pull the conversation back into reality instead of keeping it in imagined preference.

3.2 Ask less about abstract opinions and more about concrete facts

"That sounds useful," "Seems nice," and "I think I would like that" are all too abstract to guide product decisions.

Higher-value information usually looks more like this:

• I spent two hours dealing with this last week
• Right now I am holding it together with Excel and chat
• I already paid for something related to this last month
• My biggest fear is not slowness, it is making a mistake

That kind of information helps you judge the intensity of the problem, how often it happens, and whether anyone might pay to solve it.

3.3 Ask less about the user's preferred solution and pay more attention to how they solve the problem today

Users are often good at describing pain, but not always good at designing the best product.

If you ask:

• Would you want an AI to do this automatically?
• Would a smart feature help?

you usually get a vague opinion about a proposed solution, not evidence about the underlying need.

Better questions are:

• What do you do today?
• Why do you do it that way?
• What is bad about that method?

Seeing the current workaround clearly is often more valuable than asking "What do you want us to build?"

The participant, not the script, is the center of the interview. The interviewer in this photograph attends to the participant's account; the computer and paper are only recording tools. There is no need to rush through a checklist. When phrases such as “last week,” “eventually I had to,” or “I already spent” appear, follow the experience with another question. Photograph: ManonB2018 / Wikimedia Commons, CC BY-SA 4.0.

4. Why People Keep Giving Nice but Unhelpful Answers

If you understand this part, you will make fewer mistakes during interviews.

4.1 People naturally try to be polite

Especially when the person knows you, it is hard for them to say:

• this direction does not sound very strong
• I would never use this
• this is not important enough for me

They are much more likely to say something like "sounds interesting" or "could be useful."

4.2 People overestimate their future selves

Many people honestly believe their future self will:

• be more disciplined
• be more willing to learn
• be more willing to pay
• be more willing to try new tools

So the sentence "I would probably use that" often does not mean they really will.

4.3 Your question format is already shaping the answer

When you ask:

• My idea sounds pretty good, right?
• This feature would help you, right?

you are already hiding the "good answer" inside the question.

That is one reason The Mom Test strongly warns you:

do not turn the interview into a search for reassurance.

5. Weak Questions vs Better Questions

These comparisons are useful because almost every beginner asks some version of them.

Weak question	Better question
What do you think of this idea?	When was the last time this happened to you?
Would you use this if it existed?	How do you handle this now?
Would you pay for this?	Have you already spent time or money on this problem? What did you spend it on?
Is this feature important?	Which step in the process feels slowest, most frustrating, or least trustworthy?
Would you want an AI to do this automatically?	Why have you not found a better workaround yet?

The most important thing in the table is not the wording itself, but the direction of the shift:

• from opinion to fact
• from future to past
• from your solution to the user's problem

The exercise below presents six common interview statements. Decide which ones provide usable evidence and which ones are only opinions.

Comparison exercise

Distinguish behavioral evidence from subjective opinions

Judge each interview quote, then reveal the explanation.

0/ 0 correct

01

“If this tool existed, I would probably try it.”

02

“I spent an entire evening revising my résumé last week.”

03

“That sounds useful. You should build it.”

04

“I send my résumé to an alumnus now, but I feel bad repeatedly asking for help.”

05

“I think AI can definitely solve this problem.”

06

“I have paid for résumé editing services twice.”

Criterion: Has it already happened? Did the user spend time, money, or effort?

6. A Basic Interview Sequence

If you want to talk to someone now, you can use this order directly.

6.1 Open as a learner, not a seller

For example:

I am trying to understand how people actually deal with this in real life. I am not selling anything right now.

That makes it easier for the other person to drop the instinct to encourage you.

6.2 Start from the last real incident

Good opening questions are:

• When was the last time this happened?
• What happened?
• What did you do first?

Once the conversation enters one specific real event, the quality of the information usually improves a lot.

6.3 Then ask about behavior, cost, and alternatives

Continue with questions like:

• What do you do today?
• What feels worst about that method?
• How much time, money, or energy does it cost?
• Have you tried anything else? Why did you stop?

6.4 Only then judge pain and priority

You do not have to ask directly, "How painful is this?" You can often judge it from the details:

• does this happen often?
• are they already actively patching the problem?
• have they already paid some real cost?
• do they talk about it with visible frustration or emotion?

Those clues are much more useful than asking, "Is this a pain point for you?"

7. A More Complete Example

Suppose you want to build an AI product that helps college students improve resumes.

Weak questions

You ask a classmate:

I want to build an AI resume optimizer. What do you think?
If it could automatically rewrite your resume for a job description, would you use it?

They will probably say:

• sounds good
• I think that could be useful
• I would try it if it were free

Those answers give you almost no reliable signal about the actual strength of the demand.

Better questions

You can change the conversation to this:

When was the last time you edited your resume?
Why did you need to change it?
How did you do it?
Which step felt hardest?
Did you ask anyone else to review it?
Have you ever spent money or a lot of time on this?

From these questions, you may learn things like:

• many people are not bad at writing, but bad at tailoring the resume for different roles
• the biggest pain is often not formatting, but not knowing which experience belongs
• they delay not because they are lazy, but because every revision round drains them
• current workarounds already include seniors, templates, AI tools, and friends

That gets you much closer to the real problem.

8. How The Mom Test Works with JTBD

If JTBD helps you see what kind of progress the user is trying to make, The Mom Test teaches you:

how to verify through interviews whether that job is actually real.

You can combine the two like this:

1. use JTBD to draft one job hypothesis
2. use The Mom Test style questions to ask about the last real situation
3. judge whether that job is frequent, painful, and worth prioritizing

Example JTBD hypothesis:

When I am preparing internship applications, I want to adapt my old resume into a role-specific version so I can submit faster.

Now validate it with questions like:

• When was your last internship application?
• How did you edit your resume?
• Which part was hardest to rewrite?
• How did you judge whether it was ready?

That is how the two methods connect:

• JTBD helps define the need hypothesis
• The Mom Test helps validate it through conversation

Interviews help us understand what happened in the past, but they are not the final step. Once a prototype exists, users should operate it directly so we can compare what they described with what they actually do.

After “I would use it,” examine whether the solution is usable. This photograph records a 2012 Bangalore DevCamp usability test of a new translation application. A participant completes a task on the computer while researchers observe points of difficulty. Interviews test whether the problem exists; usability testing then examines whether the solution helps complete the task. Photograph: Amire80 / Wikimedia Commons, CC BY-SA 3.0.

9. Common Beginner Mistakes in Interviews

9.1 Turning the interview into a product presentation

If you explain too much about your idea, the other person starts helping you instead of telling you the truth.

9.2 Interviewing only friends

Friends are not useless, but they are more likely to encourage you. You need at least some people who are closer to real users and less emotionally invested in you.

9.3 Asking about features too early

If the problem is still unclear, detailed feature questions usually mean you are moving into solution mode too early.

9.4 Treating "I would use it" as validation

Interviews can help you judge direction, but interviews are not the whole validation step. Real validation still depends on real cost: time, switching effort, trial behavior, or payment.

9.5 Not organizing what you learned

If you do not organize the conversation afterward, it quickly becomes a blurry impression. Try to capture:

• repeated problems
• emotional words in the user's own phrasing
• current workarounds
• costs already paid
• your updated judgment

Finishing the interviews only means the raw material is ready. This affinity diagram comes from a Wikimedia Foundation usability project. Observations were first separated into individual pieces of evidence and then clustered by recurring theme. This produces a pattern across interviews rather than the impression that everyone needs the product. Photograph: Parul Vora / Wikimedia Foundation, CC BY-SA 3.0.

10. A Reusable Question Checklist

If you want to start quickly, this set is broad enough for many interviews.

Opening questions

• When was the last time this problem happened?
• What exactly happened?

Behavior questions

• How did you handle it?
• Why did you do it that way?

Cost questions

• How much time or energy does this usually cost?
• Have you ever spent money to solve it?

Alternative questions

• What other tools or methods have you tried?
• Why did you stop using them?

Closing question

• If this problem came up again, what would an ideal solution feel like?

This is fine near the end, but it should not come first. Earlier in the conversation, you want facts more than wishes.

11. Using AI to Prepare and Organize Interviews

The Mom Test is still a method for talking to real people, so AI cannot replace real interviews. But AI is extremely useful before, during, and after interviews, especially for beginners who need structure.

12.1 Rewrite weak questions

Many people know they should not ask, "What do you think of my idea?", but they still drift back to that kind of wording. You can ask AI to rewrite your draft questions first:

Below are the questions I plan to ask in user interviews:
[paste your questions]

Please rewrite them using The Mom Test principles:
1. remove opinion-based questions
2. remove future hypothetical questions
3. turn them into questions about real past behavior, current alternatives, and real costs
4. organize the result into 8-10 interview questions I can actually use

A very beginner-style input also works:

I want to ask users:
1. What do you think of my AI resume tool?
2. Would you use it?
3. Would you pay for it?

Please turn these into better interview questions.

Possible AI output:

Rewritten questions:

1. When was the last time you edited your resume?
2. Why did you need to edit it?
3. How did you do it?
4. Which part took the most time?
5. Did you ask anyone else to review it?
6. Have you ever spent money or a lot of time solving this?

That output is useful because it turns opinion-seeking questions into behavior-seeking questions.

12.2 Create different interview guides for different user types

The same problem feels different to different user groups. Students, HR people, and senior peers often care about different parts of the workflow. AI can generate separate interview guides for each group.

For example:

I want to talk to two groups:
1. college students applying for internships for the first time
2. seniors who have reviewed many resumes

Please create a 6-question interview guide for each group.

Possible AI output:

For students:
1. When was your last internship application?
2. What part felt hardest?
3. How do you judge whether your resume is ready?
...

For seniors:
1. When did you last review a junior's resume?
2. What common issues do you see most often?
3. Where do students usually get stuck?
...

That makes interview prep much easier because you do not need to invent every question from scratch.

12.3 Sort interview notes into facts vs opinions

Synthesis must remain connected to the source material. This oral-history interview uses a guide, historical photographs, and field notes together. The project description also identifies transcription as a key step in making the record usable. AI can assist with transcription and classification, but every conclusion should remain traceable to the participant's words. Photograph: U.S. National Park Service, public domain.

After interviews, the problem is often not "too little information," but "too much scattered information." AI is good at turning messy notes into structured evidence:

Below are notes from 3 user interviews.
Please organize them using The Mom Test:
1. which parts are facts and which are opinions
2. what the user's last real behavior was
3. what the current workaround is
4. what time, money, or effort cost they have already paid
5. which problems show up repeatedly
6. which statements sound positive but have weak evidence

Simple beginner input:

Here are my notes from one interview:

- she said she would probably try such a tool
- last week she spent one full evening editing her resume
- she currently asks friends for feedback
- she is not sure when a resume is "good enough"

Please separate facts from opinions.

Possible AI output:

Opinion:
- she would probably try such a tool

Facts:
- she spent one full evening editing her resume
- she currently depends on friends for feedback
- she is not sure when the resume is good enough

Useful evidence:
- the problem happened recently
- she already paid a meaningful time cost
- the current workaround depends on other people

This is especially useful because it helps beginners separate "sounds nice" from "supports a real decision."

12.4 Do a light web search before interviews

Before interviews even begin, AI can help with a light external scan:

• how people complain about the problem in public communities
• which tools get criticized most often
• whether people already spend money on related solutions
• what alternatives already exist

Example prompt:

Please look up:
"What do students complain about most when editing resumes?"
Summarize the 5 most common complaints in simple language.

Possible AI output:

Common complaints:
1. I don't know what belongs on the resume
2. I have to rewrite it for every role and it is exhausting
3. I keep editing but still do not know if it is good enough
4. I do not have reliable feedback
5. I keep delaying because I never feel ready

This does not replace real interviews, but it helps you enter them with a better starting map.

12.5 Ask AI to review your interview technique

You can also paste one interview transcript and ask AI to critique your questioning:

Here is a transcript from one user interview.
Please review it using The Mom Test:
1. Which questions sound like I was seeking reassurance?
2. Which questions were leading?
3. Where should I have asked more about facts?
4. How could I ask this better next time?

That is especially helpful for beginners because it trains the instinct to ask:

am I collecting evidence, or am I just collecting encouragement?

12. Summary

• A positive evaluation of a product idea is not sufficient evidence of demand.
• Interviews should prioritize specific past experiences over predictions of future behavior.
• Behavior, time, money, and current alternatives are the main evidence for judging problem intensity.
• AI can help prepare a guide and organize notes, but it cannot replace real interviews.

13. Exercise

🎯

动手练习

Run a User Interview Without Pitching Your Idea

1. Write 5 weak interview questions you might normally ask
2. Rewrite them in The Mom Test style
3. Interview 3 potential users about the last time the problem happened
4. Sort your notes into facts, workarounds, costs, and repeated pain points

Further Reading

• The Mom Test official site
• Rob Fitzpatrick: The Mom Test