Step One: Shut Up and Talk to Your Customers
Customer Discovery: The Startup Step Everyone Pretends to Do
Most founders pitch before they've even asked if anyone cares.
Spoiler: they don't. Not yet.
That's not because people are stupid or "don't get it." It's because you skipped the boring, un‑sexy, absolutely essential work of learning what problem they actually have. You built the movie trailer before writing the script, and now you're shocked the audience walked out.
Customer discovery isn't garnish. It's the plate the whole meal sits on. Skip it and you'll assemble a perfectly polished, gorgeously branded, meticulously animated solution to a problem no one thinks about between coffee and lunch. Then you'll post on LinkedIn about "resilience." Please don't.
Shut Up and Shovel: Excavating Pain
In interviews, you are not a salesperson. You are an archaeologist. Put the pitch deck down and pick up a shovel. Your job is to excavate pain, not to spray paint a sunset over it.
That means getting out of your own head and into the user's reality; the messy steps, the duct‑taped tools, the bit where they swear at Excel and rage‑close twelve tabs. Your only goal is to surface their words for their pain, and then shut up long enough to hear the price tag attached to it.
Start with behaviour, not opinions. "Tell me about the last time you did X." "Walk me through it step by step." "What broke? What did you swear at?"
Run the Why gauntlet (3–5 times). Pain sharpens after the third why. That's where the cost/embarrassment/inefficiency lives.
Record and tag‑team. One person listens, one person captures quotes verbatim. Future you will thank you.
Qualify hard. If your ICP is "ops managers at 5–20 site clinics", your cousin who runs a sneaker drop Discord is not an interview. He can be a friend. He cannot be data.
Bold rule: If you hear yourself pitching mid‑interview, you've already biased the answers.
Why this matters: the fastest way to ship the wrong thing is to let your pitch contaminate the data. You'll feel productive and learn nothing. Congratulations, you just built a mirror.
"But Our Market Is Everyone"
No, it isn't. "Everyone with a smartphone" is just code for we haven't done the work. A beachhead gets you speed, signal, and sales. When one small, screaming segment loves you, you get airtight messaging, cheaper acquisition, and a roadmap that writes itself.
Segment lenses: Demographic: role, company size, industry, income, age Psychographic: values/attitudes (e.g., automation‑obsessed, compliance‑anxious) Behavioural: frequency, recency, spend, willingness to pay Geographic: where they work and how that shapes the problem
Plain‑English segment example: "Automation‑obsessed ops managers at 5–20 site clinics who lose ~2 hours/day to scheduling chaos and would pay to never see Excel again."
Acid tests: * Can you name 20 of them today? * Do you know where they hang out? (Slack groups, events, subs, associations) * Can you book 10 interviews this week without cold‑spamming the planet?
Why this matters: specificity unlocks velocity. "Everyone" buys nothing. "Ops managers who hate Excel at clinics with 5–20 sites" buys Tuesdays at 10am.
Put the Figma Down
The second your cursor hovers over "Share Screen", the interview dies. Keep it anchored in their last painful attempt to solve the problem so you get receipts (behaviour), not vibes (opinions). Your job is to walk the crime scene, not read them your manifesto.
Openers that work: * "Tell me about the last time you did X." * "What did you do first? Then what?" * "What was the worst part?" * "What happens if you do nothing?" (cost/risk/delay) * "What have you already tried? Why wasn't it good enough?" * "If I gave you a magic wand, what disappears first?" * "Where do you go to find solutions?" (search terms, communities, vendors) * "Who else gets a vote on buying this?" (buyers, blockers, approvers) * "What would make this an immediate yes?" (ROI, time, guarantees)
Questions to delete from your mouth: * "Would you use this?" * "Do you like my idea?" * "How much would you pay?" (ask after proving value)
Why this matters: you're extracting buying triggers, constraints, and approval paths. That turns into ICP, messaging, pricing, and a sales motion that doesn't require miracles.
Post‑It Forensics
Your notes will look like a transcript of chaos. Perfect. Now you convert chaos into choices. The game is pattern recognition: find the pains that are frequent, severe, and funded. Everything else is interesting trivia you shouldn't build.
Highlight anything concrete (numbers, timings, dollars), emotional ("this part kills me"), or expensive (churn, overtime, delays).
One insight per sticky. Colour per interviewee. Yes, physical or FigJam.
Affinity cluster. Group rhymes; kill loners that never repeat.
Prioritise by F‑S‑W: Frequency · Severity · Willingness to pay.
Name the Job‑to‑Be‑Done in their words: "When [situation], I want to [motivation], so I can [outcome]."
Draft value prop v0.1: Who gets what result, how fast, without which current pain.
Attach receipts: quotes, screenshots, timestamps. If you can't cite a user, it's fan fiction.
Mini‑heuristic: If you can't write your promise in one sentence a user would say, you're still guessing.
Why this matters: prioritisation is product. If you don't kill loner pains and chase high F‑S‑W patterns, you'll build a museum of edge cases. Users will admire it and then go pay someone else.
MVPs: Bring Your Smoke, Bring Your Mirrors
Your pride wants polish. Your learning wants speed. Choose speed. An ugly single‑feature that hits a precise pain beats a gorgeous suite nobody needs. Wizard‑of‑Oz and Concierge aren't "cheating"; they're accelerants. You're de‑risking value before you waste months on plumbing. Your first product should be embarrassing. If it isn't, you launched too late.
Tactics that work: * Single‑feature MVP: one button, one life‑saving thing. * Wizard of Oz: promise automation; do it manually behind the curtain. * Concierge: sit with the user; push the buttons; learn 10× faster. * Landing page: "coming soon" + demos/pre‑orders > meaningless sign‑ups. * Explainer or mockups: show a value story; use it to book interviews.
Measure pull, not vibes: * Replies to emails ("when can we try it?") * Demos booked or pilots signed * Time‑to‑first‑value getting shorter * Weekly active usage (not "created an account") * "Shut up and take my money" moments you didn't nudge
Bold truth: If the only metric moving is your follower count, congrats - you launched a content strategy, not a company.
Why this matters: "pull" is the only honest signal. Vanity metrics are emotional support animals. Keep them off your roadmap.
"If you're not embarrassed by the first version, you launched too late." - Reid Hoffman
The Only Thing That Matters (Sorry, Brand Aesthetics)
You'll feel product‑market fit before you can model it: users dragging you forward, roadmap hijacked by reality, support inbox turning into a backlog of money. Until then, you are pre‑PMF. That's not an insult; it's a job description.
PMF reality checks: * Run the PMF survey: "How would you feel if you could no longer use the product?" * If ≥ 40% say "very disappointed", you're over the line. * Retention beats acquisition. Activation beats sign‑ups. * Shorten time‑to‑first‑value every sprint. * Ask weekly: "What almost made you stop using this?" Then fix that first.
"The only thing that matters is getting to product‑market fit." - Andy Rachleff
Why this matters: features don't equal fit. Only repeatable value does. Measure the stuff that proves users come back without you bribing them.
TAM/SAM/SOM Without the Fairy Tale
Acronyms are not strategy. TAM is where your investor fantasies live. SAM is the subset your current model can plausibly touch. SOM is the small, painfully real universe you could actually close with the team, time, and money you have. The bigger your SOM honesty, the faster you'll grow for real.
TAM: the fantasy kingdom (billions, dragons, universal adoption).
SAM: who you could serve with today's model and channels.
SOM: who you can close in 6–12 months with the team and cash you actually have.
Make SOM a list, not a number: * "50 paying clinics in SEQ within 6 months" is real. * If you can't name the first 10, your funnel is a mood board.
Why this matters: counting humans you can call is a growth strategy. Counting theoretical billions is a bedtime story.
Shiny Object Prevention Checklist
Shiny features feel like progress. They're not. Before you add one more toggle no one asked for, interrogate your product like a grumpy user. If you can't answer these with receipts, you're decorating a house with no plumbing.
What single problem do we solve, in a sentence a user would say?
How quickly does a new user hit first value without hand‑holding?
What's the one killer feature they would fight to keep?
What brings them back next week without a bribe?
What almost made the last user churn, and why isn't that the next task?
Guardrail: If your answer is "we'll educate the user", the product is wrong, not the user.
Why this matters: users don't want an instruction manual for your ego. They want value, fast. Everything else is optional.
A Week, A Hypothesis, A Bruised Ego
You don't need a quarter. You need seven focused days and a willingness to be embarrassed. This cadence front‑loads reality: conversations, patterns, and a tiny testable product. By next week you'll either have proof or a better hypothesis, both are wins.
Day 1: Beachhead + JTBD hypothesis: write one painfully specific segment and one job statement.
Day 2: Recruit: 10–15 qualified interviews (warm intros > cold cope).
Day 3: Interviews: run 5; record everything; no pitching.
Day 4: Synthesis: cluster insights; pick top 3 pains (F‑S‑W).
Day 5: MVP v0.1: build the tiniest thing that proves or disproves your promise.
Day 6: Expose: put it in front of 5 more; measure pull events, not likes.
Day 7: Fix and book: kill the biggest friction; book the next 10 interviews.
End‑of‑week checklist:
A segment you can say in one breath.
A value prop a user could have written.
A live, ugly thing a real person used for something that mattered.
Why this matters: repetition builds signal. One week won't make a unicorn, but it will kill guesswork.
How to Burn Money in Six Easy Steps
Think of these as the greatest hits album for burning money. When you feel yourself drifting into one, don't rationalise, interrupt. Then replace the bad habit with a calendar invite and a microphone.
Pitching mid‑interview because you got bored of listening.
Designing for "everyone" (translation: no one in particular).
Leading questions that beg for compliments.
Collecting opinions, not behaviors.
Shipping features for investors, not users.
Celebrating sign‑ups like they're revenue.
Antidote: catch yourself doing one of these → stop → open calendar → book three interviews.
Steal This Section, Then Touch Grass
You're busy. Copy‑paste these and go outside where customers live.
Interview invite DM/email: "Hey [Name], I'm researching how [role] handle [problem]. 20 mins to walk me through the last time you did it? No pitch, just learning. I'll share notes back."
JTBD sentence: "When [situation], I want to [motivation], so I can [outcome]."
Value prop v0.1: "For [very specific customer], we help [do job] in [time] so they can [measurable result], without [current pain]."
Pull metric scoreboard: * Demos booked * Pilots signed * Time‑to‑first‑value * Weekly active usage * "Take my money" moments (count them)
Why this matters: templating removes "where do I start?" as an excuse. You start here.
One Last Kick in the Shins
You don't get product‑market fit by guessing. You get it by shutting up, listening, and building ugly until people beg for more.
Now excuse me, I'm off to draw my TAM/SAM/SOM triangle, cry into a post‑it wall, and pretend my Wizard of Oz MVP isn't just me refreshing an Airtable sheet.