Product sense framework: the 6-step method for PM interviews

The 6-step product sense framework is a filter, not a script. Its job is to force you to answer two questions before you propose anything: is this a problem worth solving (viable), and would real people choose this solution over doing nothing (lovable)? In 2026, with feasibility effectively free for software products, those are the only two questions that matter. A structured answer used to differentiate candidates; now it is the baseline. What clears the bar is whether you found a real human moment and built something that earns return visits.

The steps and time budget

Total target: 22-25 minutes. This leaves buffer for follow-up questions, which Google and Perplexity now interject frequently.

Step	Time	What you are doing
1. Clarify	2 min	Narrow scope with one or two questions that would actually change your direction
2. Strategy	2 min	Name the north star metric and one real competitive gap this product addresses
3. Users	5 min	Segment, pick one specific cohort, justify the choice
4. Pain points	5 min	Map three distinct problems using different buckets
5. Solutions	5 min	Generate three options that could each fail for different reasons
6. MVP	5 min	Name the smallest thing a real user would pay for or return to weekly

The most-cited interviewer complaint across Meta, Google, and Discord is poor time management: candidates spend 10 minutes on clarify and strategy, then rush solutions into one minute and skip MVP entirely. Set a mental clock at each transition.

Step 1: Clarify (2 minutes)

Ask one or two questions that would genuinely change your direction. Not “who is the user?” (you’ll cover that in step 3) and not “what platform?” unless platform actually constrains your solution space.

Good clarifying questions address goal (growth vs. retention vs. monetization), scope (which market or use case), or a constraint that shapes feasibility. Bad clarifying questions are theatrical: they signal preparation rather than curiosity.

In 2026, clarify must address viability signals, not just build scope. At OpenAI, the deciding moment for candidates on novel-tech prompts is whether they ask penetrating questions about what is actually possible before designing. One or two surface questions followed by moving on is a failure pattern that shows up in the prototype phase.

Step 2: Strategy (2 minutes)

State the company’s north star metric and name one real competitive gap this product is meant to close. This is not a mission statement recitation; it is the anchor you return to when you cut pain points and choose your MVP.

“The north star for Discord is weekly active communicators. The gap I’m targeting is that Discord loses users who want structured, searchable knowledge bases. Everything I build should help Discord retain those users without turning into a productivity tool.”

Two minutes. Not an essay. If you cannot name the north star metric, name a proxy and state what it approximates.

Step 3: Users (5 minutes)

Segment the user base into meaningfully different groups, pick one, and justify the choice. Do not default to “casual, regular, and power users.” That segmentation is now the single most reliable signal of a weak candidate across Meta, Discord, and Google interviewers.

A Discord interviewer put it plainly: “The more specific the personas the better. Don’t just say teenagers on iPhones. And make a decision quickly. The candidates who agonized over which persona to choose signaled that they can’t prioritize.”

Segment by behavior and context, not by frequency tier or demographic label. Real segments: “first-generation college students using Discord servers as study groups,” “remote engineering managers running async standups,” “tabletop RPG communities coordinating campaigns.” Each has a distinct job and a different pain point profile.

State which segment you chose and why that segment gives you the highest leverage relative to the strategic gap you named in step 2. This is where you demonstrate prioritization, not just description.

Step 4: Pain points (5 minutes)

Identify three pain points that are genuinely distinct, not three faces of the same underlying problem. Use the four-bucket heuristic to pressure-test your list:

• Time friction: tasks that take longer than they should or interrupt flow
• Money/value gap: the outcome does not justify what the user paid in dollars, effort, or attention
• Motivational barrier: the user knows what to do but does not do it (behavior change problem, not UX problem)
• Trust and safety gap: the user is not confident the product will behave as expected, protect their data, or act in their interest

If two of your three pain points land in the same bucket, they are probably the same problem framed differently. Interviewers at well-run companies notice this immediately.

Map your chosen segment’s day or workflow at a high level before naming the pain points. Two sentences of context grounds the problems in a real human moment rather than assumptions.

Step 5: Solutions (5 minutes)

Generate three solutions that could each fail for a different reason. If all three fail under the same condition, they are all the same idea with different names.

Structure your options by scope: one near-term (quick to ship, low dependency), one platform-level (requires a new system or cross-functional dependency), one that changes the interaction model entirely. This is a heuristic for genuine diversity, not a required structure.

For AI-native companies like OpenAI and DeepMind, “we could use AI to personalize it” now rates as weak. Naming specific model constraints and capability tradeoffs rates as strong. “A small fine-tuned classifier that flags a message as needing a response, surfaced in a digest” is a different answer than “AI recommendations.”

Step 6: MVP (5 minutes)

The MVP step is a viability and lovability test, not a feature-cutting exercise. The question: what is the smallest thing a real user would pay for or return to weekly, and why does that economics work for the company?

Name two features you are explicitly cutting from your MVP and state why. This is harder than naming what you are keeping, and interviewers at Perplexity and Google now probe on this directly: “What did you decide not to build, and why?”

End with one metric that tells you whether the MVP worked. Make it falsifiable: “7-day retention rate for study-group Discord users who receive the response digest, compared to a holdout.” Not “engagement” or “active users.”

If time allows, sketch the user journey in three steps. DeepMind interviewers want you to walk all the way to solution and narrate the UX in detail; a three-step journey map at the end of the MVP section satisfies that without blowing your time budget.

Worked example: improve Discord for study groups

Clarify: “Is the goal retention of existing study-group users, or acquisition of new ones?” Retention. That narrows everything.

Strategy: North star: weekly active communicators. Gap: Discord loses students who need structured, searchable knowledge bases. This product must give them a reason to stay instead of migrating to Notion.

Users: Three segments: casual social users, gaming communities, academic study groups. I choose academic study groups because they have a specific coordination job Discord does not serve well today, and solving for them addresses the retention gap without requiring a product redesign for gaming communities.

Pain points: (1) Time friction: students cannot tell if their question was answered without scrolling through days of messages. (2) Trust gap: they are not confident important decisions made in a thread will be findable in two weeks. (3) Motivational barrier: the activation cost to post updates is high when engagement feels low and no one seems to be reading.

Solutions: (1) A daily digest that surfaces unanswered messages and new decisions, sent to members who have not visited in 48 hours. (2) A “decisions log” pinned channel that the server admin can push key messages to, creating a lightweight searchable record. (3) A commitment layer: members can mark messages as questions and the system tracks resolution, showing a “3 open questions” badge on the server icon.

MVP: Option 1. The smallest version: a plain email or push notification listing unanswered questions from the past 48 hours. Cutting: AI-summarization of long threads (add later), and the in-app resolution flow (users reply in-app instead). One metric: 7-day return rate for users who had not visited in 5 days, measured against holdout. Viability: Discord retains a churning user for near-zero marginal cost per notification.

Use it, do not recite it

Do not announce steps by name. “Now I’ll move to pain points” signals that you are running a checklist. Move naturally between stages. If the interviewer interrupts with a follow-up, engage with it directly, then return to your position without restating where you were.

Meta introduced a dedicated “Product Sense with AI” round in 2026 for senior ICs: 30 minutes of traditional product sense followed by 30 minutes of vibe-coding a prototype with internal tools. If you are interviewing at Meta at senior level, your MVP step should include a clear enough interaction model that you could build the key screen in a session. Practice narrating UI decisions, not just product decisions.

The 2026 reframe

Feasibility is effectively free for software products now. This moves weight upstream: clarify and strategy must address viability (market size, willingness to pay, business model fit), not build complexity. The pain points and users steps are where you find the genuine human moment that makes a product lovable rather than merely functional. Solutions should fail independently. And your MVP is explicitly a double test: what is the smallest thing a real user would pay for or return to, and why does that make economic sense for the company?

The framework only wins when every step filters for viable and lovable, not for structural completeness.

Strong vs. weak

Related: CIRCLES framework for the design method this extends. How AI changed PM interviews for context on why the bar shifted. Jobs to be done for the needs-identification lens that strengthens step 4.