Skip to content
PrepMint

Product Management

Product Discovery & User Research

Problem framing, interview technique, jobs-to-be-done, opportunity sizing

40 questions
Easy· 16Medium· 16Hard· 8

Last reviewed

Recommended

Product Discovery & User Research — Timed Test 2 (10 questions)

TimedEasy10 questions · 10 min
Start test

No account needed. Answers and explanations arrive when you submit.

What this topic tests

The mix every Product Discovery & User Research set is built to, and the questions published against it so far. Nothing here is hidden before you start.

Product Discovery & User Research — target difficulty mix and published question count per level
LevelTarget sharePublished
Easy40%16
Medium40%16
Hard20%8
Total40

Product Discovery & User Research — the theory

Discovery Answers Two Questions, Not One

Product discovery is usually described as "talking to users", which collapses two jobs into one. Marty Cagan splits them. First, you need to discover whether there are real users out there that want this product. That is the first. And Second, you need to discover a product solution to this problem that is usable, useful, and feasible. That is the second. A team can finish the second and fail the first, and the result is a well-built answer to a problem nobody has.

The Four Risks Every Idea Carries

Cagan names four, and they are the vocabulary most product interviews assume you already have. Value risk asks whether customers will buy it or users will choose to use it. Usability risk asks whether users can figure out how to use it. Feasibility risk asks whether our engineers can build what we need with the time, skills and technology we have. Business viability risk asks whether this solution also works for the various aspects of our business — legal, sales, support, finance, brand.

The four are independent. Evidence that settles one says nothing about the other three.

What to Answer Before the Work Starts

Cagan's opportunity assessment is a short list of questions asked before a team commits, and it doubles as a structure for the "how would you evaluate this idea?" interview prompt:

  • Exactly what problem will this solve? (value proposition)
  • For whom do we solve that problem? (target market)
  • How big is the opportunity? (market size)
  • What alternatives are out there? (competitive landscape)
  • Why now? (market window)
  • How will we measure success/make money from this product? (metrics/revenue strategy)

Two things are easy to miss. The list demands answers before the effort rather than after it, and it treats Why now? (market window) as first-class — an idea can be right and three years early.

A Worked Pass: One Idea Through Both Filters

A team building invoicing software for freelancers proposes in-app chat with clients. Run the four risks first. A usability session shows every participant finding and sending a message without help, so the usability risk looks answered — and nothing else does. Value is untouched: nobody has shown a freelancer would choose it over the email thread they have. Feasibility is untouched: no engineer has costed message retention or search. Viability is untouched: support has not said who answers a client's complaint sent through your product at 2am.

Now the assessment. The team can answer the problem and the target market. It cannot answer why now, and it cannot say how success would be measured. Two of six unanswered is not a verdict, but it is the honest state of the idea.

Opportunities Are Needs, Not Features

Teresa Torres's opportunity solution tree is the most reproduced artefact in modern discovery, and the term it defines is the one teams get wrong. In this context, an opportunity is an unmet customer need, pain point, or desire. It is not a feature request. The tree is Opportunity solution trees are a simple way of visually representing the paths you might take to reach a desired outcome. The desired outcome sits at the top; beneath it sit the opportunities, which are These are the customer needs, pain points, and desires that, if addressed, will drive your desired outcome.; solutions hang off those. You do not start from a backlog and reverse-engineer a need to justify it.

Who Does Discovery

Torres's answer is the product trio. A product trio is typically comprised of a product manager, a designer, and a software engineer. The point is not headcount. When a product trio works together to develop a shared understanding of their customer, they are in a much better position to create products that customers love. The shared understanding is the deliverable, and it is the thing a circulated research report cannot produce.

Interviews About Last Tuesday

The most useful interview technique is also the easiest to state. Keep the interview grounded in specific instances of past behavior. Torres's own opening is Tell me about the last time you watched Netflix. — not "do you watch Netflix?", and certainly not "would you use a feature that…".

A question about the future returns a prediction, and predictions about your own behaviour are cheap. A question about last Tuesday returns a story, with the context, the constraints and the workaround still attached.

Test the Assumption, Not the Idea

When a prototype of a whole concept fails, a team learns that something is wrong. It rarely learns what. Torres's alternative is deliberately narrower: Assumption testing makes it clear that we’re testing a single assumption and not the whole idea. The payoff is diagnostic — When an assumption test fails, we know exactly what needs to change. A test that can only return "the idea did not land" leaves you nowhere in the morning.

Five Users, and What That Finding Actually Says

Almost every product manager can quote this one and almost nobody quotes both halves of it. The Nielsen Norman Group result is that The best results come from testing no more than 5 users and running as many small tests as you can afford. The advice that travels with it is that it is better to distribute your budget for user testing across many small tests instead of blowing everything on a single, elaborate study

The number is only defensible with the second half attached. Quoted alone, "five users is enough" becomes a licence to test once and stop — the opposite of what the finding says.

The Job, Not the Customer

Jobs to Be Done is a lens, not a segmentation scheme. The Christensen Institute defines it as Jobs to Be Done is a lens that reveals the circumstances—or forces—that drive people and organizations toward and away from decisions. Its central claim is that People don’t simply buy or pick products or services; they pull them into their lives to make progress.

The practical consequence is what you segment by. Two people of the same age, income and job title, in different circumstances are not the same customer.

Needs, and the Things That Only Look Like Needs

The UK Government's Service Manual is unusually blunt about evidence, and its definitions travel well beyond government. ‘User needs’ are the needs that a user has of a service, and which that service must satisfy for the user to get the right outcome for them. Note what that excludes: a need is a need of the service, not a wish in general, and it is tied to an outcome the user is trying to reach.

Then the sentence that settles most stakeholder arguments: Treat any opinions or suggestions that do not come from users as assumptions that have to be proven by doing research. An executive's strongly held suggestion is not a requirement. It is a hypothesis with a sponsor, and it enters the backlog on the same evidence as anything else.

Research Is a Habit, Not a Phase

The same manual is equally direct about scheduling. This means doing small batches of user research in every iteration of each development phase - starting in discovery and continuing throughout live. And the number teams flinch at: All team members should watch real users interacting with your service and talking about it - ideally for at least 2 hours every 6 weeks.

Who you research with is not a detail either. For your research to be effective, your participants must be actual or likely users of your service. — a colleague's opinion of a screen is not a finding. Avoid using your own staff for research into the public facing parts of your service., because Their knowledge and experience can mean they will use the service in a very different way. And screening deserves care for a reason that is easy to miss: Some people with access needs do not identify as having a disability and it is important to be aware of this in the screening process.

Sources

Marty Cagan / Silicon Valley Product Group · Teresa Torres / Product Talk · Nielsen Norman Group · the Clayton Christensen Institute · the UK Government Digital Service's Service Standard and Service Manual · the 2020 Scrum Guide. Every quotation above is verbatim, machine-verified against the live pages on 2026-08-24.

Sample questions

Three questions from this topic, with the answer and the reasoning shown.

Q1EasyA director sends the team a list of changes he is certain users want. Under the Service Manual, what is that list?
  • A set of requirements, because of the seniority of the person sending it
  • A set of needs, once the team has agreed the changes make sense
  • A set of assumptions, to be proven by doing researchCorrect
  • A set of findings, equal in weight to what the analytics reports

Explanation

The principle — what an input counts as depends on where it came from, not on how sure its author is.

Why the key is correct — the instruction is to Treat any opinions or suggestions that do not come from users as assumptions that have to be proven by doing research. A director is not a user, so the list is hypotheses. Not a dismissal: many may be right, and research finds out which.

Why the others are wrong — seniority changes who must be persuaded, not what was claimed. Team agreement converts nothing: the standard names research, not consensus. And a record of what people did is different evidence from a prediction.

Remember this — if it did not come from a user, it is an assumption with a sponsor.

Sources — the UK Government Digital Service Manual on learning about users and their needs.

Open this question on its own page

Q2EasyAmong Cagan's four big risks, which question does value risk ask?
  • Whether the product can be built and sold at an acceptable margin
  • Whether the evidence so far has raised the team's overall confidence
  • Whether customers will buy it or users will choose to use itCorrect
  • Whether users can figure out how to use it

Explanation

The principle — each of the four risks asks one question, and value asks about choice.

Why the key is correct — Cagan's value risk is whether customers will buy it or users will choose to use it It is about the decision a person makes when nothing compels them to make it.

Why the others are wrong — margin belongs to business viability, which Cagan keeps separate. General confidence is not a risk at all; naming four questions is what replaces it. And whether someone can work out the interface is usability, which is answered by watching them try, whereas value is answered by watching them decide.

Remember this — value is about choosing, not about understanding, affording or building.

Sources — Marty Cagan on the four big risks, Silicon Valley Product Group.

Open this question on its own page

Q3EasyCagan's opportunity assessment is a short list of questions a team answers before it commits to building. Which pair opens it?
  • What problem this solves, and who has already agreed to buy it
  • What problem this solves, and for whom that problem is solvedCorrect
  • What the team built, and what it learned once the thing had shipped
  • What problem this solves, and what the launch metrics later showed

Explanation

The principle — the assessment is a set of questions asked before effort is spent, not a report written after it.

Why the key is correct — Cagan's list opens with Exactly what problem will this solve? (value proposition) and For whom do we solve that problem? (target market) A team that cannot say what problem it is solving, and for whom, has nothing the rest of the list can build on.

Why the others are wrong — a signed-up buyer is a later commercial fact. Describing what was built turns the exercise into a retrospective, which is the one thing it is not. And launch metrics arrive after the decision the assessment exists to inform.

Remember this — the problem and the person come first, and both are answerable before anything is built.

Sources — Marty Cagan on assessing product opportunities, Silicon Valley Product Group.

Open this question on its own page

Practise all 40 questions

Every published question in Product Discovery & User Research, with its answer and explanation.

Frequently asked

What people ask about product discovery & user research.

Is testing with five users really enough?
Only if you quote the whole finding. The Nielsen Norman Group result is that The best results come from testing no more than 5 users and running as many small tests as you can afford. Both halves are load-bearing. The five is a claim about one round, not about a research programme, and the advice that ships with it is that it is better to distribute your budget for user testing across many small tests instead of blowing everything on a single, elaborate study A team that runs one five-person study and stops has used the number against itself. In an interview, saying "five per round, several rounds" is the answer; saying "five users is enough" is the trap.
What is the difference between product discovery and user research?
User research is a method; discovery is the decision the method feeds. Cagan frames discovery as two findings rather than one activity: First, you need to discover whether there are real users out there that want this product. And Second, you need to discover a product solution to this problem that is usable, useful, and feasible. Research is how a team gets evidence for either. The Service Manual treats research as a rhythm rather than a stage, since This means doing small batches of user research in every iteration of each development phase - starting in discovery and continuing throughout live. A team can run continuous research and still not be doing discovery, if none of it changes what gets built.
A stakeholder is certain about what users want. How do I push back without a fight?
Reclassify rather than argue. The Service Manual's line is the one to borrow: Treat any opinions or suggestions that do not come from users as assumptions that have to be proven by doing research. That is not a rejection; it moves the suggestion from settled to testable, which is a much easier conversation. It also gives you a definition to point at, since ‘User needs’ are the needs that a user has of a service, and which that service must satisfy for the user to get the right outcome for them. A senior stakeholder's instinct is often a good hypothesis. It is simply not yet evidence.
What actually counts as an opportunity, as opposed to a feature request?
Teresa Torres draws the line by definition. In this context, an opportunity is an unmet customer need, pain point, or desire. A feature request is a proposed solution, and putting it where an opportunity belongs hides the need it was meant to serve. On an opportunity solution tree — Opportunity solution trees are a simple way of visually representing the paths you might take to reach a desired outcome. — solutions hang beneath opportunities rather than replacing them. The test is simple: if you can only describe it by naming the thing you would build, it is a solution, and the opportunity underneath it has not been articulated yet.
Why am I told not to ask users what they want?
Because the answer is a prediction, and Torres's technique is built to avoid collecting predictions. Her rule for a story-based interview is to Keep the interview grounded in specific instances of past behavior. The demonstration question is Tell me about the last time you watched Netflix. — a request for one specific occasion rather than a general policy. What comes back is a story, and a story carries the context, the workaround and the constraint that a stated preference leaves out. It is also checkable in a way that an opinion about a hypothetical feature is not.