Rob Snyder, author of The Power of Pull and writer of The Physics of Startups newsletter, shares a pragmatic, field-tested framework for finding product-market fit (PMF) that he has used to help over 50 startups go from zero to $1M+ ARR. The core insight: PMF is not about convincing people to buy; it is about discovering people who are already trying to solve a problem right now and cannot do it with their current options — people who will “rip the product out of your hands.”
The Pull Framework: What Causes Someone to Buy Instantly
Pull describes a person who would be weird not to buy — a coiled spring trying to get a specific project done now but blocked by the limitations of every existing option.
The framework is an acronym (PULL) that must be fully present for real pull to exist; missing any element means the prospect will complain but not buy.
P – Project: A concrete item on the person’s to-do list (a job to be done), not a vague desire. Example: “Minimize time spent taking client-meeting notes and entering them into the CRM.”
U – Unavoidable: The project must be done right now for a specific reason (regulatory deadline, boss’s mandate, legal risk). Example: Financial advisors must file compliant notes or risk SEC trouble.
L – List of Options: The alternatives the person has already considered (build in-house, hire a consultant, use a category leader, generic tools).
L – Limitations: Why every option on that list fails to get the project done (wrong integrations, non-compliant, too slow, too expensive).
Pull is a person-level property, not a company-level one. You don’t sell to “SpaceX”; you sell to the Head of Legal at SpaceX who has a stuck project on her personal to-do list.
When pull is real, the conversation sounds like: “Wait, you can do that? How do we start? Does the product exist yet?” — not polite enthusiasm or “send me a deck.”
Finding an Idea Worth Building: Earned Secrets Over Brainstorming
The most reliable path to a pull-backed idea is firsthand immersion — “getting your hands dirty” alongside the people you want to serve — because these secrets don’t emerge from whiteboards or interviews alone.
Rough breakdown of idea origins across 50+ startups: ~10% founder’s own lived experience, ~90% deep, in-person shadowing/working with target users (sometimes even buying a franchise to see the P&L).
Example: A founder with no idea shadowed solar installers (hated it), then physical therapists (loved it), watched them for days, tried to buy a franchise, and spotted a specific cost-cutting project they were stuck on.
Pull Hypothesis: Distill the observed stuck moment into a two-slide test:
“Here’s what I hear: you’re trying to do X right now because Y, you’ve looked at [options] but they fall short because Z. Is that roughly right?”
“If so, we could help by [tiny conceptual sketch]. Does that fit what you’re looking for?”
Signal of truth: The person says “Oh my god, yes” to slide 1 and “How does it work? How do we move forward?” to slide 2 — ideally trying to pay you before a product exists.
Anti-patterns to avoid:
Building first (“I’ll show them a demo”) — delays the cold wind of reality.
Design partners — they give visionary feedback but never buy; real pull buyers don’t want to be design partners, they want the job done.
Solving your own problem only if you would have genuinely paid someone else to solve it; otherwise it’s a “should exist” fantasy.
Landing the First Sales Calls: Founder Magic Outreach
Goal: 2–3 conversations/day with potential buyers. Start cold if no warm intros.
Message template: Short, human, interesting — not a value-prop pitch. “Hey, Snoop Dogg invested in us and I thought you’d find it interesting. Open to chat?” or “I dropped out of my PhD to build for content marketers. Thought you might find it interesting.”
Benchmarks on LinkedIn: 30–40% connection acceptance → 30–50% reply rate → majority say “sure, why not.” If below, rewrite the “interesting” hook or fix your profile.
No AI personalization — it reads like Claude and adds zero signal. Low personalization is fine if the hook is genuinely interesting to that persona.
The message only gets the meeting; pull either exists on their to-do list or it doesn’t — nothing you say in the DM creates it.
Running the Sales Call: Discovery First, Pitch Tiny
Vibe: Relaxed, “let’s see if I can help,” not “convince you to buy.” If pull exists, they will pull; if not, no amount of pushing changes that.
Structure (30-min calendar block):
“Thanks for the time. Curious — you get lots of outreach, why take this call?” → They often fill out the whole PULL framework for you.
If they say “You reached out, tell me,” use their answer as the agenda: “Great, we’ll cover that. Sound good?” → Get thumbs-up.
“Before we jump in, mind if I ask a few questions to understand your context?” → Earns the right to ask discovery questions without feeling like an interrogation.
Discovery questions map to PULL: “Is there anything around [area] you’re trying to do now but can’t with current tools/team?” → “Why does this have to happen now?” → “What else have you looked at? Is it good enough?”
Green flag: Lean-in urgency — “How do we start? What’s the price? I’ll pay right now.”
Red flag: Lean-out enthusiasm — “Love the vision, send a deck, we’ll review in Q4.”
Demo: Usually not on call 1 (too early). If you have one, keep it tiny and only after pull is confirmed.
Stop rule: ~5 calls with zero pull → pause, review recordings, ask the Hail Mary (see below), or pivot hypothesis.
The Hail Mary & Goblin Mode: When Pull Isn’t There
Hail Mary question (asked at end of a lukewarm call): “Is there anything on your to-do list right now that you’d pay someone to take off your plate?”
Multiple startups pivoted to their winning idea from this single question (e.g., gov-tech ATS → job-description automation).
Goblin Mode (break-glass tactic): Do whatever unhealthy/crazy thing gets you in front of buyers now — fly to a conference tomorrow, parents door-knock dental offices, schedule 20 meetings in a day. Goal: find anyone who will buy right now to generate a new hypothesis.
Iterate: New hypothesis → 5 calls → assess → repeat until pull repeats across multiple people.
Closing the First Deal: Fit Then Project Management
Fit conversation: Come back in 2–3 days with a rough mockup/vibe-coded prototype. “Based on our talk, here’s what I think fits. What else do you need to know to decide?”
Pricing talk track: Anchor to the value of the project or cost of current alternatives (the L’s). “Others have paid $80k–$1M depending on scope; how are you thinking about it?” — lets them react to a case-study range instead of you naming a number.
No pilots, no POCs, no design partnerships — they convert “I need this now” into “I’m helping you build.” If procurement requires a pilot, comply, but never offer proactively.
Enterprise project management: Ask the champion, “When does this need to show results?” Work backward from that date to map every required step (security review, legal, demos) — creates a forcing function.
Early delivery often service-shaped: Humans do the work behind the scenes while automation catches up; validates pull and teaches what to build.
Post-Sale: Retention Is a Leading Indicator You Must Design For
Default: Every new customer is guaranteed to churn until they hit X — the moment it would be weird for them to leave.
X is rarely usage/NPS. Examples:
Slack: 10k messages sent in first X days.
Restaurant retention software: First prevented-quit notification within two weeks.
Action: Deliver firsthand (don’t throw a login over the fence). Keep asking “Are you going to churn now?” until you see the “I can’t live without this” moment, then build a bullet-train onboarding that guarantees every customer hits X.
If mass churn: Don’t guess — find the 1–2 segments not churning. One horizontal AI workflow startup discovered freight forwarders used one specific workflow; pivoted entire company to that premium vertical.
Scaling: The Case Study Factory Mental Model
A startup is a factory that repeats one hell-yes case study infinitely.
The case study is the PULL story: “Customer had Project P, Unavoidable U, Options L failed because L → we delivered Z → they said hell yes before buying and after using.”
Three factory stations = bottlenecks to attack one at a time (Theory of Constraints):
Pipeline: Getting the right people to talk to us.
Sales: Turning conversations into closed deals.
Delivery: Getting customers to the “weird to churn” moment.
Sales becomes: “Here’s the case study we deliver. Does this look like you? If not, we’re probably a bad fit.” — enables qualification and disqualification.
Founders (or product leaders at mature cos) must run this loop themselves for 0→1; delegating too early loses the earned secret.
Key Principles & Trap Summary
Pull > Push: If you feel like you’re convincing, something is wrong — either your hypothesis or your target.
All four PULL elements must co-exist; partial pull = polite “no.”
Earned secrets come from reality, not theory — shadow, work, live the workflow.
Sell to learn, not to close — every call either confirms pull or teaches you why it’s absent.
No design partners, pilots, or free POCs — they attract the wrong signal.
Retention is engineered, not hoped for — find the leading indicator X and build the bullet train to it.
Tarpit idea: Chasing a space with a clear winner without a specific unblocked-person theory (e.g., “Linear won with better UX, so I’ll build better-UX HR software” — but HR buyers aren’t blocked by UX).
Metal detector vs. spoon: This framework is a metal detector on a beach; you still have to walk the sand, but you’re not sifting grain by grain with a spoon.
