A persona project typically ends one of two ways: it produces a plausible-looking fictional character no one ever references again, or it produces a persona genuinely extracted from real customers. The first version doesn't help anyone make marketing decisions.
This article is the interview question library for the second outcome — six categories of questions that pull a persona out of real conversations. Plus the reasons you should not let AI build your persona on your behalf.
What personas are actually for
A persona is "the shared language for discussing marketing strategy at individual-level resolution":
- "Target = women in their 30s, office workers" — no one can picture anyone
- "Target = Sara (35, IT firm, working mom, up at 5 AM)" — everyone pictures the same person
So the purpose is aligning the team on a shared decision premise. Not "guessing the right person."
That said, if the persona is wildly off from real customers, the premise of the discussion is wrong — which is why the persona needs to be extracted from real interviews, not built from imagination.
The six categories
Category 1: Attributes and lifestyle
Understand their background.
1. "Tell me about your current role."
2. "What's your function and standing in your team?"
3. "Walk me through a typical weekday schedule."
4. "What do weekends look like?"
5. "What's your household situation, and your role at home?"
6. "Has your lifestyle changed much in the last few years?"
7. "Anything you're aiming to do in the next 1–3 years?"
Design notes
- Capture the person across time (past → present → future)
- Ask about life-stage changes
- Get the balance between work, home, and hobbies
Category 2: Behavior patterns
What they actually do, in time-and-place specifics.
1. "Tell me about a recent purchase that stuck with you."
2. "How does your behavior differ between weekdays and weekends?"
3. "When in your life do you feel time-starved?"
4. "Conversely, when do you feel time-rich?"
5. "Anything you've started or stopped recently?"
6. "What's the 'I never skip this' habit?"
Design notes
- Pay attention to what they keep doing — that's where the essence shows
- Pay attention to what they stopped — that's the past version of them
- Probe their relationship to time
Category 3: Values and priorities
"What do you care about?" The hardest category.
1. "What matters most to you at work?"
2. "What do you spend on without hesitation? What do you never buy no matter how cheap?"
3. "If you had to save either time or money — which? Does it depend?"
4. "Have you had a recent 'I'm not budging on this' moment?"
5. "When do you feel your values don't match other people's?"
6. "If no one were watching, what would you do?"
7. "Any values you'd want to pass on to a kid or someone you mentor?"
Design notes
- Ask about concrete choices, not abstract values
- "Spend on without thinking" and "never buy" are honest signal
- "If no one were watching" strips the public-facing layer
Category 4: Media and information diet
Where they get information from.
1. "What's a recent piece of industry news you noticed? Where did you see it?"
2. "What media, social platforms, or sites do you read regularly?"
3. "How do you gather information for work?"
4. "When you're buying something, where do you start researching?"
5. "Whose information do you trust? Whose do you distrust?"
6. "Has anyone's content changed your behavior recently?"
Design notes
- Beyond what they consume, ask what they trust
- Tie "where they heard about it" back to purchase decisions
- Map out influencer / expert contact patterns
Category 5: Purchase decision process
Reconstruct the process of an actual decision.
1. "Walk me through a recent big purchase, from consideration to closing."
2. "What alternatives did you compare?"
3. "What was the deciding factor at the end?"
4. "Did you talk to anyone before deciding?"
5. "Looking back, was it a good or bad decision?"
6. "If you had to do it again, what would you change?"
Design notes
- Time-sequenced reconstruction
- Ask the breadth of alternatives considered
- Identify the decision maker (self, family, manager?)
Category 6: Obstacles, anxieties, stress
Surface unmet needs and worries.
1. "What's been bothering you lately?"
2. "When do you feel everyday stress?"
3. "What 'if only X existed' wishes do you have?"
4. "Anything about how you spend your time or money that you regret?"
5. "What are you anxious about over the next 3–12 months?"
6. "If you had one day with no constraints, what would you do?"
Design notes
- "What's frustrating?" — usually weaker than asking about stress and regret
- Hypothetical framing like "if you had one day" surfaces real motivation
- Anchor in time horizons (3 months, 1 year) to draw out future-self imagery
From interview to persona
Step 1: Interview multiple people
Minimum 5, ideally 8–10. Build diversity in.
Step 2: Transcribe and extract striking lines
Pull memorable quotes, behaviors, values from each transcript.
Step 3: Find shared patterns
Themes that show up across multiple respondents:
- Common anxieties
- Common decision criteria
- Common information sources
- Common values
Step 4: Write the persona
A working persona format:
■ Name (placeholder): Sara Mitchell
■ Profile: 35, PMM at an IT firm, $90k salary, lives in a major metro,
married with a 4-year-old daughter
■ Weekday: Up at 5 AM, prep deck in the morning, podcast on commute,
daycare pickup after work
■ Values: Efficiency first; willing to pay to buy back time;
prefers "real" things but is overwhelmed by info
■ Sources: Industry newsletter, X (industry accounts), word of mouth from trusted friends
■ Purchases: Decides herself, but talks to her partner;
strong "don't want to be wrong" pull
■ Obstacles: Time scarcity, choice overload,
wants "real" but can't tell signal from noise
■ Our role: "The partner that saves time and decision cost — and is trustworthy."
Step 5: Socialize and operate
- Everyone on the team should be able to recall the persona without prompting
- In decisions, ask: "What would Sara do?"
- Refresh the persona every 6 months
The hard part — why you shouldn't build the persona with AI
"Ask ChatGPT to build a persona" is in fashion. It has structural problems.
Problem 1: AI returns the average
AI is trained on public data, so it returns the typical person in the target category. That's an average, not a real individual.
Problem 2: Your customer's particulars don't surface
Asking AI to build "a persona of women in their 30s" never returns what's specific to your customer base. AI hasn't seen your first-party data.
Problem 3: No team dialogue
When you build a persona through interviews, your team directly engages with real customers. That's how marketing instincts develop. AI-generated personas skip the development entirely.
Problem 4: No internal credibility
"This persona is AI-generated" inspires zero buy-in. A persona that comes from 5–10 real interviews is something the rest of the org will actually take seriously.
The right use of AI
Don't use AI to be the persona. Use AI for:
- Brainstorming the interview script
- First-pass theme extraction from transcripts
- Synthesizing multiple interview outputs into structure
- Formatting the final persona document
AI as work accelerator, humans as interpreters of real customers — that's the AI-era division of labor.
Persona shelf life
Personas are not a one-time deliverable.
- Refresh every 6–12 months
- Update earlier when the market or customer base shifts
- Re-check the relevant persona on every new feature / new business decision
- Be willing to retire personas that no one is using
"A persona built three years ago, gathering dust in someone's drawer" is effectively the same as having no persona at all.
Where Repoan fits
Repoan supports persona work around the interviews:
- Screening surveys — recruit diverse interview participants
- AI-generated attribute questions — script design support
- Cross-customer open-text comparison — discover persona seeds via theme extraction
- Continuous attribute capture — keep personas current over time
Summary
Interview questions for building real personas:
- Six categories (attributes, behavior, values, media, purchase, obstacles)
- Use concrete choices and hypothetical framing to pull honest answers
- Synthesize from 5–10 interviews into shared patterns
- Don't outsource persona creation to AI (use AI for support work only)
- Persona is an ongoing program, not a deliverable
A persona is not a fictional ideal — it's a shared image extracted from real customers. The "extracted-from-real-customers" property is what makes the persona useful for marketing decisions.