Customer Discovery for Founders
Stop Building
in the Dark
Know exactly who to talk to, what to ask, and how to spot real demand before you spend months building the wrong thing.
Most founders build what people say they want. Not what they’ll actually pay for.
The external problem
You have an idea you believe in, but no reliable way to test it before you’ve already spent months building.
The internal problem
Everyone you pitch says “that sounds great.” You can’t tell if they mean it. So you keep building on encouragement that isn’t proof.
What's at stake
Six months of your life. A launch into silence. The founders who win tested demand before they built.
Four steps. Clear output at each one.
Use the full flow if you’re starting cold, or jump straight to the step that matches where you are now.
Strategy & Targeting
Define who you’re actually looking for, which assumptions could kill the idea, and what problem they’re most urgently trying to solve.
Outreach
Write honest, low-friction messages that get real people to say yes to a conversation, not flag you as another cold pitch.
The Interview
Run a conversation that pulls out real stories, workarounds, budget signals, and what people actually do, not what they say they’d do.
Analysis & Signals
Turn messy notes into patterns. Rank real problems. Decide whether to dig deeper, pivot, or start testing a solution.
Reference library: Not advice. Tested frameworks.
Each step is grounded in proven customer discovery methodology. The Mom Test teaches you what questions not to ask. JTBD tells you what to listen for. Lean Startup tells you what to do with what you find.
Supporting reference files follow the practical guide. Archived source material and author workflow notes are not a current policy certification.
01 Strategy & TargetingMap risky assumptions and lock in the job-to-be-done. Lean Startup + JTBD.
Reference
Strategy & Targeting
The mindset
Customer discovery is a detective exercise. You're confirming or denying your assumptions. You're not selling, not pitching, not asking customers to design your product. You're there to learn.
Fall in love with the problem, not the solution.
Every business is a stack of assumptions. Your job is to test the high-impact, high-uncertainty ones before investing heavily.
You're an intelligent filter, not an order-taker. Discovery informs decisions, it doesn't make them. Vision still leads.
Don't talk to anyone with a pulse. Target the specific customer you think has the problem.
Everyone on the team should talk to customers. Not just the designer or the founder.
The 4 reasons companies buy
Every B2B purchase maps to one of these:
- Make money
- Save money
- Gain competitive advantage
- Avoid compliance or risk
Map every pain point you hear to one of these. If it doesn't tie to one, the purchase is unlikely.
Map your assumptions first
Answer these before you interview anyone:
- Who is my target customer?
- What problem do they want to solve?
- Why can't they solve it today?
- What measurable outcome do they want?
- Who is my earliest adopter?
- How will I acquire customers?
- How will I make money?
- Who is my primary competition and why will I beat them?
- What assumption, if proven wrong, would kill this?
Mark the ones that are both high-impact and highly uncertain. Those are your interview priorities.
Define your customer segment
"Everyone" is not a customer. Get specific: job, role, mindset, context.
Not "marketing managers." Try "marketing managers at 50–200 person B2B SaaS companies who own the content pipeline and report to the VP of Marketing."
Pick one narrow segment and one specific job to solve first.
Define three audiences:
- Typical customer at scale (who you'll serve once you're big)
- Early adopter (feels the pain acutely, loves trying new things, interview first)
- Critical distribution or fulfillment partners
B2B buyer landscape
Map every role involved in the purchase:
- User: who actually uses the product
- Buyer: who signs the contract
- Payer: who controls the budget
- Technical buyer: who evaluates fit
- Economic buyer: who approves the spend
- Strategic buyer: who owns the outcome
- Influencer / recommender: who shapes the decision
Start with mid-level managers, not C-suite. Easier access, more conversations, and you'll be better educated before going up the chain. Influencers and recommenders are especially valuable because they connect you to more customers.
Jobs-to-be-done framing
Define the job as the specific problem they're trying to solve in a specific situation.
Include three layers:
- Functional needs (what they need to get done)
- Emotional needs (how they want to feel)
- Social needs (how they want to be seen)
Describe the problem in the customer's own words, not founder language. Pull phrasing from forums, Reddit, and actual conversations.
Validating the problem
The problem needs to be:
- Important to the customer
- Unsatisfied by current options
- High-value enough that they'd pay for a better solution
If any of the three fails, it's not worth solving first.
Identify what they do today instead of your solution. Existing alternatives matter. DIY workarounds are strong signals of real need.
Distinguish primary from secondary problems. Solve the pain that matters most.
Finding interview subjects
Three rules:
- One degree of separation. Don't interview close friends or family.
- Be creative. Don't expect people to come to you.
- Fish where the fish are. If a tactic isn't working, switch immediately.
Tactics:
- End every interview with a goal of 2–3 new referrals
- LinkedIn search by title, reach out via referral or cold
- Call the CEO's office, ask the EA for "the person who handles X." They route you. Then name-drop the CEO's office in follow-up.
- Find people at their moment of pain (conferences, relevant forums, events)
- At conferences: ask for time after, not during. Follow up while memory is fresh.
- You're a student. Say so. It's an advantage. Offer to share research results.
- Start with warm intros before cold outreach. Always exhaust that first.
Default to asking for advice early on. "I'm not selling anything, I just want 20 minutes." People like being asked. It makes them feel important.
No surveys. Period.
Don't use surveys. Not for discovery, not for recruiting, not for screening.
Why:
- You need the customer's exact words, context, workflow, and emotions
- Early-stage discovery is about uncovering the real problem, not confirming your assumptions
- Conversations let you probe, clarify, and discover unexpected needs
- Surveys flatten nuance and kill follow-up
- They're noisy and usually poorly designed
- Stronger proof points come from hearing pain, dissatisfaction with current alternatives, and actual purchase intent directly
The right path:
- Use quick research or LLMs to learn the space
- Do primary research through live customer conversations
- Document pain points and proof points in their own words
02 OutreachLow-friction messages that get people to a conversation. Converted at 10%.
Reference
Outreach
Warm intros first. Always.
A mutual connection asking on your behalf will always outperform cold. Exhaust warm intros before sending a single cold message.
If you're not finding people, it's rarely "I don't know how." It's usually fear of rejection in disguise. Send more messages. Faster.
The 2-sentence formula
Two sentences. That's it.
Sentence 1: Their job to be done.
Lead with a specific problem they likely deal with. Not your idea. Not your product. Make it relatable enough that they instantly recognize it. Hint at the problem, don't explain your solution.
Sentence 2: The honest ask.
Say exactly who you are and what you want. Ask for a specific time, not "sometime next week."
No-oriented close: "Would it be crazy to ask for 15 minutes tomorrow at 3pm?"
On LinkedIn you have 300 characters. Use every technique that fits.
Example outreach
Simple version:
I'm working to better understand how software sales leaders handle pricing when customer value varies so much across deals. Could you do a 15-minute call tomorrow at 6pm to share how you approach it?
Stacked version (LinkedIn note, 300 chars):
Hi Kendall. I'm a student at Purdue researching marketing workflows at B2B companies for a concept competition. Exploring AI's impact on team workflows.
I ONLY need 10 interviews by Friday to move on to the next phase!
Would it be crazy to ask for 15 minutes of your time?
- Elliot
This converted at 10%. 1 in 10 cold strangers agreed to a call.
The 12 tricks that make it work
- Lead with their world. Open with a problem they recognize, not a pitch they have to decode.
- Zero threat. "Student" + "concept competition" signals nothing is for sale. Guard drops immediately.
- Complete honesty. Be upfront about who you are and what you want. People respond to realness.
- Tiny ask. 15 minutes doesn't feel like a decision.
- Specific time. "Tomorrow at 6pm" creates urgency. "Sometime next week" gets ignored.
- Scarcity. A cap + deadline makes them feel like a limited slot, not one of a thousand targets.
- Hard deadline. "By Friday" forces action now. Later always becomes never.
- Visual interrupt. Capitalizing ONLY pulls the eye straight to the scarcity line.
- No-oriented close. Saying no means agreeing it's crazy. Path of least resistance is yes.
- Curiosity gap. Specific enough to be relevant, vague enough that they want to know more.
- Human format. First-name opener, personal sign-off. No links, no title.
- No pitch. No ask to buy, no demo, no product link. Pure curiosity.
What to avoid
- Long paragraphs
- Leading with features or your product concept
- Multiple value props crammed in
- Vague asks like "sometime next week"
- Anything misleading about who you are
- Framing that looks like sales
Keep testing
Test lots of versions. Tone, opening line, specific time, deadline. Small tweaks move conversion more than you'd expect.
03 Interview ExecutionPull out real stories and workarounds, not speculation. The Mom Test.
Reference
Interview Execution
Prepare an interview guide
Always write one. It's a guide, not a script. Veer off when something interesting comes up.
Structure:
- Brief intro explaining purpose
- 1–2 warm-up questions
- Story-eliciting behavioral questions
- Drill-down follow-ups
- Close: "What should I have asked you that I didn't?"
Set pass/fail metric targets before you start. Know what you're hoping to see. Write it down.
Don't wing it. No prepared, prioritized questions = wasted conversation.
Don't try to learn everything in one session. Go deep on a few areas instead.
Format priorities
In-person is best. Video is second. Phone is last resort. Never text or email.
One person at a time. No focus groups. Group think contaminates answers.
Bring a note taker if possible. Stay present during the conversation. If you're solo, write notes immediately after. Detail fades fast.
The most important rule: stories, not speculation
Humans are terrible at predicting their own behavior. "Would you use this?" and "Would you buy this?" answers are weak on their own.
Ask about the past instead:
- "Tell me about the last time you..."
- "Walk me through how you..."
- "What triggered your search for a solution?"
- "What happened after you tried it?"
Past behavior is the strongest predictor of future behavior.
Ask open-ended questions
Start with: who, what, why, how.
Avoid: is, are, would, do you.
Don't rush to fill silence. Let them sit. Gold comes out in the pause.
Drill down relentlessly
Whenever something tweaks your antenna, go deeper. Multiple layers of "why."
Parrot back what they said. They'll either correct your misinterpretation or refine their own thinking.
Look for solution hacks
If someone has DIY-solved the problem or cobbled together a workaround, that's one of the strongest market-need signals you can get. Nobody builds workarounds for problems they don't feel.
Test for priority
Even if the problem exists, it needs to be acute enough to change behavior and justify payment.
Ask where it ranks in their list of problems. Ask what they're spending time on this week vs. this. If the pain isn't top-5, it's probably not getting solved.
Pricing questions
Speculative pricing is unreliable. "Would you pay $X?" answers mean almost nothing on their own.
Better:
- Ask what they currently spend to address this problem
- Ask what budget exists and who controls it
- Create situations where they think they're actually buying. People only honestly consider willingness to pay when it feels like a real transaction.
Handling prototypes
If you're showing something:
- Behavior questions first, product feedback second. People love talking features and it'll contaminate their stories.
- Explicitly ask for brutal honesty upfront
- Don't trust verbal "I like it." Watch behavior. Try to get them to open their wallet.
Disarm your own bias
Go in ready to hear things you don't want to hear.
Adopt a "try to kill the idea" mindset for interviews. It works.
A dedicated note-taker helps. Interviewers filter for what they believe. Note-takers catch what you miss.
Don't lead the witness. Don't embed the answer in how you ask.
Don't pitch. Don't hear only what you want to hear. Don't treat one conversation as truth.
Close every interview with this
"What should I have asked you that I didn't?"
Then: "Who else should I talk to?"
Goal: 2–3 referrals per conversation.
04 Analysis & SignalsTurn interview notes into ranked patterns and next-step decisions.
Reference
Analysis & Signals
How to take notes
Top of every note doc:
- Subject name
- Date
- Interviewer(s)
- Format (in-person, video, phone)
- Photo if possible
- Brief descriptive info about the subject
Capture direct quotes and the story. Not your interpretations.
Convert notes to metrics
Calculate percentages across interviews. Numbers fight cognitive bias.
Don't obsess over single numbers. Don't treat qualitative results as statistical fact. Keep questioning what's behind the numbers.
Dump and sort exercise
Team exercise, 10 minutes:
- Everyone writes patterns they noticed on sticky notes
- Put them on a wall
- Sort into groups
- Discuss what clusters and what doesn't
- Re-review assumptions. See what needs to change.
Do this every 5–10 interviews.
Pattern recognition thresholds
A clear pattern across 5–10 people is a signal.
Even 5 consistent responses can reveal a valid pattern if the responses are tight.
Don't overreact to one conversation. Don't wait for statistical significance either. Move fast from credible patterns. Analysis paralysis is the enemy.
Stickiness as a signal
The strongest signal isn't "they liked it." It's whether they'd be genuinely upset if the product disappeared.
If you could yank it away and they wouldn't care, you don't have stickiness yet. Keep digging for what would actually hurt them to lose.
Don't stop at "they liked it." Identify the specific value they got and why they care.
Progression signals to watch for
Signal gets stronger at each step:
- They describe the pain clearly and unprompted
- They say they'd buy a solution
- They ask how to buy it now
Level 3 is gold. Level 2 counts. Level 1 alone isn't enough.
Also strong:
- They've DIY-hacked their own solution
- They're currently paying for a partial workaround
- They can tell you their existing budget and who controls it
The Truth Curve
Each level yields more reliable truth, at higher cost:
- Conversations
- Paper test (landing page, deck, concept test)
- Prototype
- Live product (real behavior, real money)
Start at the cheap end. Build confidence iteratively. Concierge or Wizard-of-Oz experiments (manually faking the product) beat interviews because you watch actual behavior.
Interview count target
Goal: 10–20 interviews.
Patterns often show up fast. Even 5 consistent responses can reveal a signal. Top student teams typically talk to 20–30.
If you hit 20 without patterns emerging, your segment or hypothesis is probably off. Rethink before interviewing more.
Primary vs secondary problems
Not every problem you hear is the one you should solve.
After interviews, force-rank the problems by:
- How often they came up
- How intense the emotion was
- How willing people were to act on them
- How much money or time they were already spending
Solve the #1 problem. Don't get distracted by #3.
Before vs after lens
For each problem you might solve, sketch:
- What does the customer's world look like today?
- What does it look like after your solution exists?
- What specifically changes: cost, revenue, time, trust, peace of mind, compliance risk?
Quantify wherever possible. Is your solution 1% better, 50% better, 3x better, 10x better? Aim for multiples, not percentages.
Real user vs real decision-maker
These are often different people.
The user feels the pain. The decision-maker signs off.
Map both. Interview both if you can. A product users love but economic buyers won't approve doesn't ship.
Confidence vs impact framework
After patterns emerge, list every assumption left.
Rank them on two axes:
- How confident are you? (low to high)
- How badly does the business hurt if you're wrong? (low to high)
Test low-confidence, high-impact first. Those are your biggest risks.
Apply this across the whole model: customer segment, value prop, channels, revenue, costs, differentiation, success metrics.
False positives
People want to be nice. Weight positive reactions with skepticism.
"I love it" means nothing without behavior to back it up.
Watch what they do, not what they say.
When to transition from discovery to selling
Trigger: you start hearing the exact same thing, interview after interview.
That's pattern saturation. Shift from discovery to offering a solution. Start testing whether it's burning pain they'll pay to solve or mild discomfort they'll live with.
Don't wait for a perfect sample size. Hit the trigger, move.
You're never fully done
Customer discovery is ongoing. Qualitative research is permanent in good companies.
When patterns flatten, evolve what you're exploring. The questions change. The cadence shouldn't stop.
Worst thing you can do: stop talking to customers after week one.
Use it now. It's free.
Works on all platforms. Pick yours and get set up in under a minute.
Claude.ai
Download & upload in 60 seconds
Download the zip, then open Claude.ai to upload it.
Download the zip below. With JavaScript enabled, Claude.ai opens in a new tab if your browser allows it. Otherwise, use Open Claude.ai.
Click the + button in the skills column on the left.
Select Create a skill.
Select Upload a skill and upload the zip.
Start your customer discovery session by running /estack-customer-discovery.
Claude Code
Run one command in any terminal
Works anywhere Node.js is installed — your project directory, home directory, anywhere.
Open your terminal app and run this command.
npx elliot-stack@latest
Run the command.
Start Claude Code by typing claude in your terminal.
Start your customer discovery session by running /estack-customer-discovery.
ChatGPT / Other
Paste a prompt — the AI does the rest
This prompt tells the AI where to fetch current instructions from the live site. No installs. No version lag.
Copy the prompt below.
Preview prompt
# Customer Discovery This skill guides users through the full customer discovery process, or any specific part of it. Customer discovery is a detective exercise — you're confirming or denying assumptions, not selling or pitching. ## The flow There are 4 steps. Each builds on the last, but the user can jump to any step directly. 1. **Strategy & Targeting** — Define who you're talking to, what you're testing, and why. Map assumptions, define customer segments, frame the job-to-be-done. 2. **Outreach** — Craft messages that get people to say yes to a conversation. Warm intros first, then cold outreach using proven formulas. 3. **Interview Execution** — Build a tailored interview guide with the right questions (behavioral, story-eliciting) and avoid the wrong ones (speculative, leading). Prepare to run great interviews. 4. **Analysis & Signals** — Make sense of what you heard. Spot patterns, rank problems, identify progression signals, and decide what to do next. ## How to use this skill ### Step 1: Figure out what the user needs Ask the user: do they want to go through the full flow, or do they need help with a specific step? If they describe their situation without picking a step, route them based on where they are: - Have an idea but haven't talked to anyone yet → Start at **Step 1: Strategy & Targeting** - Know who to talk to but need help reaching out → **Step 2: Outreach** - Have interviews scheduled but need questions → **Step 3: Interview Execution** - Have done interviews and need to make sense of them → **Step 4: Analysis & Signals** ### Step 2: Read the right step file — MANDATORY <CRITICAL> You MUST read the step file AND its corresponding reference file BEFORE responding to the user. Do NOT skip this. These files contain the actual frameworks and instructions — without reading them, your output will be generic and wrong. </CRITICAL> Each step has its own workflow file with detailed instructions. **Read the step file first, then read the reference file it points to.** Both reads are required — not optional, not "if you have time." | Step | Step file (read first) | Reference file (read second) | |---|---|---| | 1. Strategy & Targeting | `https://elliotdrel.com/e-stack/customer-discovery/steps/step-1-strategy.md` | `https://elliotdrel.com/e-stack/customer-discovery/references/01-strategy-and-targeting.md` | | 2. Outreach | `https://elliotdrel.com/e-stack/customer-discovery/steps/step-2-outreach.md` | `https://elliotdrel.com/e-stack/customer-discovery/references/02-outreach.md` | | 3. Interview Execution | `https://elliotdrel.com/e-stack/customer-discovery/steps/step-3-interviews.md` | `https://elliotdrel.com/e-stack/customer-discovery/references/03-interview-execution.md` | | 4. Analysis & Signals | `https://elliotdrel.com/e-stack/customer-discovery/steps/step-4-analysis.md` | `https://elliotdrel.com/e-stack/customer-discovery/references/04-analysis-and-signals.md` | The reference docs are the source of truth — they contain the principles and frameworks. The step files tell you how to work through them with the user. ### Step 3: Be adaptive Read the room. If the user clearly knows what they're doing, don't over-explain. If they're new to this, coach them. Adjust your depth and pacing based on how they talk about their idea. ### Step 4: Offer deliverables At the end of each step, offer to produce a written summary the user can reference later. Don't force it — some users just want the conversation. ### Step 5: Continue the flow After completing a step, offer to continue to the next one by name (e.g., "Would you like to continue to **Step 3: Interview Execution**?"). Never reference a step by number alone — always include the name so the user knows what it covers. If the user jumped into a specific step, still offer the next step — they might want to keep going. --- ## Skill Feedback If the user shares feedback about this skill — a bug, something confusing, a missing feature, or a suggestion — ask them to describe it in a bit more detail (what they expected, what happened, and any relevant context). Then file the issue using whichever method is available: **If `gh` is installed** (`gh --version` succeeds), create the issue directly: ```bash gh issue create \ --repo ElliotDrel/e-stack \ --title "estack-customer-discovery: <concise summary>" \ --body "<description from user feedback — expected vs. actual behavior and context>" ``` **If `gh` is not installed**, build a pre-filled URL: ```bash python3 -c " import urllib.parse title = 'estack-customer-discovery: <concise summary>' body = '<description from user feedback — expected vs. actual behavior and context>' base = 'https://github.com/ElliotDrel/e-stack/issues/new' print(base + '?title=' + urllib.parse.quote(title) + '&body=' + urllib.parse.quote(body)) " ``` Share the printed URL with the user and offer to open it in their browser. They can also click it directly, review the pre-filled title and body, and click **Submit new issue**.
Open your platform and paste the prompt.
Start your customer discovery session.
Customer Discovery ships in E-Stack, a set of 21 free skills installed by one command. See the whole stack.