Watch what a founder does in the half-second after you ask a good question. They start toward the real answer, the specific one, the Tuesday in March when the deal fell apart on a pricing call. Then they catch themselves and reach for the safe version. "We learned a lot about our pricing that quarter."
That half-second is the entire job.
Story extraction gets sold as a list of clever questions. It isn't. Any founder will give you serviceable answers to a good prompt. What separates a session that fills a content bank from an hour you'll never get back is what you do in the moment a founder skips the real story and substitutes the rehearsed one. This piece covers both: the questions that open the door, and the listening discipline that's the actual work.
The questions are the door, not the room
The questions are a script for getting a founder past their investor answers. They are not the extraction. Here are the seven we use, phrased as they get asked.
- Who is your ICP and what do they actually care about? The word that matters is "actually." Most founders have a rehearsed ICP answer written for a board deck. The real answer lives one layer below: what the buyer says on a Tuesday call when they drop the pitch voice.
- What's changing in your space that most people aren't paying attention to? Not the trend everyone's writing about. The shift the founder notices from their seat that hasn't made it into the discourse yet. The best answers start with "I've been noticing" and come at half speed.
- What's the hardest part of your stage right now? The first answer is a board-update version. The second is closer. The third is usually the one worth keeping.
- Tell me about the team around you. What makes them special? Team stories get skipped because founders worry they read as self-congratulatory. They don't, when the specifics are there: a named person, a decision, a moment someone did something the founder couldn't.
- What's one thing you'd do differently if you were starting over? Regret reads as self-awareness, which is the scarcest signal in a founder feed.
- What do most founders in your space get wrong? A contrarian question, asked carefully. The useful version names a specific assumption the founder disagrees with, not a sweeping dismissal of the category.
- How do you do things differently than competitors? About operational choice, not feature list. What the founder's team does on a Tuesday that a competitor's doesn't. This is where the company's personality shows up.
Every founder story sits in one of four buckets: a pattern, a tension, a risk, or a belief. Questions 1 and 2 surface patterns. Question 3 surfaces tension, where most unfiltered material lives. Questions 4 and 5 surface risk. Questions 6 and 7 surface belief. A call that covers all four produces roughly five to ten story seeds, enough for a month of posts and a sales narrative that sounds like the founder.
Now the part nobody writes down.
The four signals that a real story just went by
A founder almost never refuses to tell you the real story. They start to, then route around it. The skill is hearing the route happen in real time. There are four signals, and once you're listening for them you can't stop.
| Signal | What it sounds like | What it usually sits on top of |
|---|---|---|
| The half-second hesitation | A beat of silence, then a more general answer than the question deserved | A specific moment they decided wasn't safe to say out loud |
| The "anyway" bridge | "...anyway, the point is..." | The thing they almost said before the "anyway" |
| The minimizer "just" | "It was just a small thing, but..." | A story they've pre-judged as not worth your time |
| The abandoned specific | They name a person, a date, a number, then move on without it | A full scene attached to that detail |
The "anyway" bridge is the most reliable, because it's involuntary. When a founder says "we almost lost the account, anyway the point is our onboarding got tighter," the story isn't the tighter onboarding. It's the almost-lost account, handed to you and taken back in the same breath. The half-second hesitation works the same way: when they land on "scaling the team has been a challenge," the pause told you a sharper answer got set down. And the abandoned specific is the most wasted of the four. A founder drops a real detail ("our biggest churn was a Series B fintech in Q2") and keeps moving as if it were context. It wasn't context. It was a scene they walked past.
What to do in the half-second after they skip it
When you hear a signal, you do one thing: stop, and send them back to the exact spot. Quietly, without making it an interrogation. The move is to repeat their own abandoned words back as a question. Not "tell me more," which is a content-strategist tic founders can feel coming. Use their language.
- They say "anyway." You say: "Wait, go back. You almost lost the account."
- They say "it was just a small thing." You say: "What was the small thing?"
- They drop a date and move on. You say: "March. What happened in March?"
- They hesitate, then generalize. You say: "You paused before you answered that one. What was the version you didn't say?"
That last one feels risky and almost always works, because it names what the founder already knows happened. You're not extracting anything. You're giving them permission to say the thing they were already half-saying. Most founders are relieved someone noticed.
The discipline is restraint. You're not steering toward a topic you want; you're following the founder to the thing they flinched away from, because the flinch is the signal that it's real. A session where you got clean, complete answers to all seven questions usually produced nothing usable. Clean answers are rehearsed ones.
Record before you announce, not after
Turn the recording on before you tell the founder you've turned it on. The setup matters more than any single question.
This isn't a trick. It's about where the good material lives. A founder's most usable language shows up in the warm-up, when they think the real conversation hasn't started. The moment you say "okay, let me hit record," the register shifts. They sit up and start performing the founder-on-a-podcast version of themselves, the exact rehearsed layer you're trying to get under.
So: get permission to record at the top, start recording immediately, then have the loose pre-conversation as if it's off the clock. Tell them at a natural beat that you've been recording and you'll cut anything they want. Founders rarely ask to cut the warm-up. That's where they sound like themselves.
Grounded questions beat the origin myth
Ask about the last 90 days, not the founding story. The origin myth is the most rehearsed material a founder owns, and the least useful.
Every founder has told their origin story a hundred times. It's smooth, it carries the fundraise, and it's been sanded down to nothing specific. Ask for it and you get the press-kit version. The seven questions above are grounded in the present tense for this reason: the hardest part of your stage right now, what's changing today.
Recent material is unrehearsed material. A founder hasn't had time to polish a story from three weeks ago into an anecdote. The pricing call that went sideways last Tuesday still has the friction on it: the specific numbers, the thing the customer actually said. That friction is what makes a post sound like a person instead of a brand. The craft of running the session in the room lives in the interview playbook.
The Upshot
Every founder is sitting on more story than they can see, too close to it and trained to route around the specific in favor of the safe. The shortage is never material. It's hearing.
For the operator: learn the questions once and stop thinking about them. Put your attention on the four signals, and when you hear one, send the founder back to the exact spot in their own words. That single habit is the difference between a transcript full of brand statements and a content bank full of stories that sound like the person who lived them.
For the founder doing this on yourself: the questions work alone with a recorder running, but the listening doesn't. You can't catch yourself skipping your own stories in real time, because you're the one skipping them. Get a second person to ask and listen. That's the mechanism, not a nice-to-have. And if your stories keep coming out generic, the problem usually starts before the writing, with why founder stories sound generic.
Extraction is the first step. Shaping and deploying the material are a different craft, covered end to end in the founder communications guide. But none of it happens without the raw stories, and the raw stories don't come from better questions. They come from better listening.
Common questions.
What is story extraction?
Story extraction is the practice of pulling real, specific story material out of a founder through a recorded conversation, then turning it into content. It is a listening discipline, not a question list. The questions open the door. The skill is hearing when a founder skips past the real story and stopping them before it's gone. In my experience, a 60-minute session run well produces roughly five to ten story seeds.
What questions do you ask to extract a founder's story?
Seven questions cover the ground: who your ICP actually cares about, what's changing in your space that most people miss, the hardest part of your current stage, the team around you, what you'd do differently, what most founders in your space get wrong, and how you do things differently than competitors. Each maps to one of four story types: pattern, tension, risk, belief. The questions are a starting script, not the work itself.
How do you know when a founder is telling a real story versus a rehearsed one?
Four audible signals. A half-second hesitation before they pick the safe version. The word 'anyway' used to bridge past something they almost said. The minimizer 'just' ('it was just a small thing'), which usually sits on top of the real story. And an abandoned specific, where they name a person, a date, or a number and then move on without it. Each one is a place to stop and ask them to go back.
Can a founder run story extraction on themselves?
Partially. The questions work alone with a recorder running, but the listening discipline does not. The four signals are nearly impossible to catch in your own speech in real time, because you're the one skipping past the story. If you can't bring in a content operator, enlist a colleague or friend to ask the questions and listen while you record. A second person hearing the answers live is what makes it work.
