| name | four-steps |
| description | Applies Steve Blank's Customer Development methodology from The Four Steps to the Epiphany. Use when a startup is searching for customers and a business model before scaling. Covers the four-step process: Customer Discovery (find if anyone wants what you're building), Customer Validation (prove you can sell it repeatably), Customer Creation (drive demand matched to Market Type), and Company Building (transition from learning org to execution org). Triggers include 'we built it but no one's buying', 'should we hire salespeople yet', 'how do we find our first customers', 'we're burning cash and sales aren't scaling', 'are we in a new or existing market', 'when do we scale'. NOT for companies that have already crossed the chasm into mainstream (use Crossing the Chasm instead), not for optimizing an existing sales funnel, not for product development methodology (this is its companion, not replacement). |
Note: This skill is independent analysis and commentary, not a reproduction of the original text. It synthesizes the book's core ideas with modern startup practice, surfaces where frameworks are outdated or incomplete, and integrates perspectives from adjacent disciplines. For the full argument and context, read the original book.
The Four Steps to the Epiphany
"Startups that survive the first few tough years do not follow the traditional product-centric launch model. Through trial and error, hiring and firing, successful startups all invent a parallel process to product development. I call this process 'Customer Development.'" - Steve Blank
Should You Use This Skill?
Are you searching for customers and a business model (not executing a known one)?
├─ YES → This skill is for you
└─ NO → You have paying customers and a repeatable sales process?
├─ YES, but stuck before mainstream → Use Crossing the Chasm
├─ YES, scaling fine → You're past this. Use Company Building concepts only
└─ NO, still pre-product → Start here at Customer Discovery
Have you validated that people will pay for your product?
├─ NO → Start at Customer Discovery (Step 1)
└─ YES → Can you describe a repeatable sales process?
├─ NO → You're in Customer Validation (Step 2)
└─ YES → Have you matched your launch to your Market Type?
├─ NO → You're in Customer Creation (Step 3)
└─ YES → Company Building (Step 4)
The Core Insight
9 out of 10 new product introductions fail. Not because the products don't work - because the companies never found customers.
The Product Development model (Concept → Build → Test → Launch) was designed for manufacturing in the early 1900s and adopted by consumer packaged goods in the 1950s. It works when you're launching a new product into an existing market with known customers. Most startups have neither.
PRODUCT DEVELOPMENT (what everyone does):
Concept/Seed ──→ Product Dev ──→ Alpha/Beta ──→ Launch
↑
"Where are the
customers?"
CUSTOMER DEVELOPMENT (what winners do - in parallel):
┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
│ Customer │←──→│ Customer │←──→│ Customer │───→│ Company │
│ Discovery│ │Validation│ │ Creation │ │ Building │
└──────────┘ └──────────┘ └──────────┘ └──────────┘
↻ ↻ ↻
iterate iterate iterate
Customer Development is a companion to Product Development, not a replacement. Each step is an iterative circle - going backwards is natural and valuable, not failure.
"It's OK to screw it up if you plan to learn from it." - the heart of the methodology.
The Ten Flaws of the Product Development Model
| # | Flaw | Consequence |
|---|
| 1 | Where are the customers? | Greatest risk is customer/market development, not product development |
| 2 | Focus on first customer ship date | FCS means Engineering is done, not that you understand customers |
| 3 | Execution instead of learning | Sales/marketing hired for what they know, not what they can learn |
| 4 | No meaningful sales/marketing milestones | "Hire, fire, repeat" substitutes for real progress metrics |
| 5 | Product dev used to measure sales | Clock to tell temperature - measures the wrong thing |
| 6 | Product dev used to measure marketing | All plans made in "a vacuum of real customer feedback" |
| 7 | Premature scaling | Fully staffing sales/marketing before knowing if anyone will buy |
| 8 | Death spiral | Premature scaling → burn rate → missed numbers → fire VP Sales → fire VP Marketing → fire CEO |
| 9 | Not all startups are alike | Four Market Types need radically different strategies |
| 10 | Unrealistic expectations | "Build it and the customers will come" is not a strategy, it's a prayer |
"In a startup, no facts exist inside the building, only opinions."
The Four Market Types
Everything depends on Market Type. It changes how you find customers, how you launch, how you spend, and how long it takes to become profitable.
| Market Type | Definition | Competitors | Time to Profit | Key Risk |
|---|
| Existing | Higher performance than what's currently offered | Known incumbents define the market | 12-18 months | Cost of entry (Lanchester rules) |
| New | Enables something customers couldn't do before | Non-consumption / other startups | 3-7 years | Market may never materialize |
| Resegmented (Low-cost) | "Good enough" at substantially lower price | Incumbents who abandon low margins | 18-36 months | Must be profitable at low price |
| Resegmented (Niche) | Radical enough to change the rules for a subset | Incumbents who defend profitable core | 18-36 months | Segmentation must be spot-on |
"Market Type changes everything a company does."
How to Tell Which Type You're In
Are there existing companies selling comparable products?
├─ NO → NEW MARKET
│ (No competitors, but also no existing customers)
└─ YES → Are you competing on the same basis (features, performance)?
├─ YES → EXISTING MARKET
│ (You must be better on known dimensions)
└─ NO → Are you targeting a subset of customers?
├─ YES, at lower price → RESEGMENTED (LOW-COST)
└─ YES, for a specific need → RESEGMENTED (NICHE)
The Handspring/Palm Test
Same team, same CEO (Donna Dubinsky), same product category, three years apart:
- Handspring (1999, existing market): PDA market was billion-dollar. Differentiated on expandability. $170M revenue in 12 months. Customers knew what a PDA was.
- Palm (1996, new market): PDA market didn't exist. If Palm had used Handspring's "drive demand from competitors" playbook: zero dollars in sales. Had to educate customers about what a PDA could do.
Identical products and team. Wrong Market Type strategy = death.
Step 1: Customer Discovery
"The goal of Customer Discovery is finding out who the customers for your product are and whether the problem you believe you are solving is important to them."
What You Are NOT Doing
- Collecting feature lists from prospects
- Running focus groups to define the product
- Writing a Marketing Requirements Document
- Understanding the needs of ALL customers
What You ARE Doing
- Testing whether your founders' vision matches a real, painful problem
- Building the product iteratively for the few, not the many
- Getting outside the building
"The initial product specification comes from the founders' vision, not the sum of a set of focus groups."
Earlyvangelists - The Most Important Customers You'll Ever Know
Not all early customers are equal. You need earlyvangelists - visionary customers who:
| # | Characteristic | Why It Matters |
|---|
| 1 | Has a problem | Not latent - they know it exists |
| 2 | Understands the problem | Can articulate it and its cost |
| 3 | Actively searching for a solution | Has a timetable |
| 4 | Has cobbled together an interim solution | Proves the pain is real enough to act on |
| 5 | Has or can acquire budget | Can actually buy |
The customer pain hierarchy:
"What problem?" → Latent need. Useless for 2 years.
"Yes, terrible. I hand out water." → Active, but only papering symptoms.
"Losing $500K/yr. Need X by Y." → Visualized solution. Getting warm.
"Wrote a req to IT, they cobbled → HAS A HOMEMADE SOLUTION. YES.
something but it keeps crashing."
"Budgeted $500K for a vendor." → HAS BUDGET. ULTIMATE CUSTOMER.
The Four Phases of Customer Discovery
| Phase | Activity | Key Question |
|---|
| 0 | Get Buy-In | Does the team agree to this process? |
| 1 | State Your Hypotheses | What do we believe about customer, problem, product, pricing, channel, market, competition? |
| 2 | Test the Problem | Do customers recognize and care about this problem? |
| 3 | Test the Product | Does our product concept solve their problem? Will they pay? |
| 4 | Verify | Iterate or proceed to Customer Validation? |
Six hypotheses to articulate before leaving the building:
- Customer & Problem Hypothesis
- Product Hypothesis
- Distribution & Pricing Hypothesis
- Demand Creation Hypothesis
- Market Type Hypothesis
- Competitive Hypothesis
Pass/Fail
"If, and only if, you are successful in this step do you proceed to Customer Validation."
Passing means: founders' vision matches a real, painful, paid-for problem. Failing means: iterate within Discovery or exit.
Step 2: Customer Validation
"Customer Validation is where the rubber meets the road."
The Goal
Build a repeatable sales roadmap - "the playbook of the proven and repeatable sales process that has been field-tested by successfully selling the product to early customers."
What You Are NOT Doing
- Staffing a sales team
- Executing a sales plan
- Executing a "sales strategy"
- Generating revenue (that's a side effect, not the goal)
"Building a roadmap to sales success, rather than building a sales organization, is the heart of Customer Validation."
Sales Roadmap vs. Sales Pipeline
A pipeline is the funnel (leads → suspects → prospects → closes → orders). Mature companies need it.
A roadmap answers: Who influences? Who recommends? Who decides? Who has budget? Who sabotages? How many calls per sale? Average cycle? What selling strategy? Key customer problems? Profile of the optimal earlyvangelist?
"It is impossible to build a sales pipeline without first having developed a sales roadmap."
Core Questions Customer Validation Must Answer
- Do we have product/market fit?
- Do we understand the sales process?
- Is the sales process repeatable?
- Can we prove it's repeatable? (Proof = full-price orders)
- Can we get orders with the current product?
- Have we positioned correctly?
- Do we have a workable channel?
- Can we scale profitably?
The Validation Team
Founders/CEO must be in front of customers through at least the first iteration. Delegating to VP of Sales is often fatal.
- In enterprise/B2B, if no founder can close: hire a "Sales Closer" (NOT a VP of Sales)
- Sales Closer profile: regional-manager background, great Rolodex, comfortable with ambiguity, not interested in building an org
Pass/Fail - The Canonical Pivot
Three conditions to pass:
- Repeatable customers
- Repeatable sales process
- Profitable business model
"If you can't find enough paying customers in the Customer Validation step, the model returns you to Customer Discovery to rediscover what customers want and will pay for."
This is the pivot - predating the term's popularization. Failure here is not failure; it's learning. Most startups cycle through Steps 1-2 at least twice.
Step 3: Customer Creation
"Customer Creation builds on the success of the company's initial sales. Its goal is to create end-user demand and drive that demand into the company's sales channel."
Why It Comes AFTER Validation
"No serious spending in marketing until the company has a proven and repeatable sales roadmap."
Cash protection: this step is placed after Validation to move heavy marketing spending after the point where a startup has its first customers.
The Fatal Error
Most startups execute the same launch playbook regardless of Market Type. This is the #1 Customer Creation mistake.
Customer Creation Strategy by Market Type
| Building Block | Existing Market | New Market | Resegmented Market |
|---|
| Year 1 Goal | Market share | Market adoption | Market share + segment education |
| Positioning | Differentiation (faster, cheaper, better) | Vision and passion ("what could be") | Segmentation (unique valued spot) |
| Launch Type | Onslaught (full frontal) | Early Adopter (targeted, low-cost) | Niche (focused onslaught) |
| Demand Creation | Drive demand into channel | Educate about the market | Educate + drive demand |
New Lanchester Strategy (Market Entry Cost)
Military operations research applied to market entry:
| Leader's Market Share | Implication | Cost of Entry |
|---|
| ≥74% (monopoly) | Do NOT attack head-on | 3× their budget - unaffordable |
| ≥41% (clear leader) | Very difficult | Resegment or create new market |
| 26-41% (unstable) | Some opportunity | 1.7× weakest player's budget |
| <26% (fragmented) | Market ripe for entry | Affordable entry possible |
"Your goal is to become No. 1 in something important to your customer."
First Mover Advantage Is a Myth
Golder & Tellis (1993), 500 brands in 50 categories:
- 47% of market pioneers failed
- Early market leaders (entered ~13 years later on average) had 8% failure rate
"The issue is not being first to market, but understanding the type of market your company is going to enter."
The Three Launch Types
Onslaught (Existing Market): Full-frontal assault. Maximum exposure. Heavy upfront spending. Only correct for capturing share in an existing market.
Early Adopter (New Market): Targeted, low-cost. Goal is mind share, not market share. Education campaign targeting earlyvangelists. Create a "tipping point." New markets take 3-7 years to become profitable.
Niche (Resegmented Market): All demand-creation dollars focused on one identifiable segment. If the segment is speculative, treat as new market instead.
Messengers (from Gladwell's "Law of the Few")
Three types to educate before launch:
- Experts - industry analysts, product reviewers, consultants. Value independence.
- Evangelists - paying customers who are unabashedly enthusiastic. "You can't get them off the phone."
- Connectors - bloggers, conference organizers, thought leaders who bridge multiple worlds.
Phase Summary
| Phase | Activity |
|---|
| 1 | Get Ready to Launch - choose Market Type, set year-1 goals |
| 2 | Position Company and Product - audits, match positioning to Market Type |
| 3 | Launch - select launch type, audiences, messengers, craft sticky messages, measure |
| 4 | Create Demand - match demand strategy to year-1 objectives, measure, iterate |
Step 4: Company Building
"Company Building is where the company transitions from its informal, learning and discovery-oriented Customer Development team into formal departments with VPs of Sales, Marketing and Business Development."
The Two Failure Modes
- Entrepreneur refuses to adapt: insists "more of the same" chaos works at scale
- Investors replace founders with "professionals": bureaucracy kills innovation DNA
Both are wrong. The answer is a mission-centric organization - the third alternative between startup chaos and corporate rigidity.
Three Stages of Company Evolution
| Stage | Organizational Form | Activities |
|---|
| 1 (Discovery/Validation) | Customer + Product Development teams | Learning and discovery |
| 2 (Company Building) | Mission-centric organization | Cross the chasm |
| 3 (Large company) | Process-centric organization | Repeatable/scalable execution |
The Four Phases
| Phase | Activity |
|---|
| 1 | Reach mainstream customers - match growth curve to Market Type |
| 2 | Review management, build mission-centric organization |
| 3 | Transition Customer Development team into functional departments |
| 4 | Build fast-response departments (OODA loops) |
Mainstream Customers by Market Type
- Existing market: Linear ramp. No real chasm. Use positioning to differentiate. Relentless execution.
- New market: Hockey stick. Long flat period. Husband resources, evangelize, grow market. Risk: market may be a chimera.
- Resegmented market: Hybrid. Some early sales mask the need to convert mainstream. Use branding AND positioning.
Mission-Centric Management (from USMC Warfighting Doctrine)
Five components:
- Mission Intention - every mission has tasks AND intention. Intention is more enduring.
- Employee Initiative - taking initiative is part of the implicit employment contract.
- Mutual Trust - good news fast, bad news faster. Never punish the messenger.
- "Good Enough" Decision-Making - "A good plan violently executed now is better than a perfect plan next week." (Patton)
- Mission Synchronization - peer-wise cross-departmental coordination, not top-down staff meetings.
When Founders Should Stay
Blank rejects the "kick the founder out" reflex:
- Within 3 years every company faces a competitive innovation challenge
- The corporate DNA for innovation that founders carry is precisely what's needed
- Many startups die after founders are ejected because process-oriented replacements gut innovation
"The long-term success of a startup requires founder continuity long past the point when conventional wisdom says the founders should be replaced."
Decision Trees
"Which step are we in?"
Do you know who your customers are and what problem you solve?
├─ NO → CUSTOMER DISCOVERY (Step 1)
└─ YES → Can you sell repeatably at a profit?
├─ NO → Have you sold to anyone?
│ ├─ NO → Still CUSTOMER DISCOVERY
│ └─ YES → CUSTOMER VALIDATION (Step 2)
└─ YES → Are you driving demand and growing?
├─ NO → CUSTOMER CREATION (Step 3)
└─ YES → COMPANY BUILDING (Step 4)
"Should we hire salespeople?"
Do you have a proven, repeatable sales roadmap?
├─ NO → DO NOT HIRE SALESPEOPLE
│ Founders sell. Maybe one Sales Closer.
└─ YES → Are you past Customer Validation?
├─ NO → Still NO. Iterate first.
└─ YES → Now you can staff a sales org.
"What Market Type are we?"
Can customers already do what your product does (via competitors)?
├─ NO → NEW MARKET
│ • Year 1 goal: adoption, not share
│ • Launch: early adopter (low-cost, targeted)
│ • Timeline: 3-7 years to profit
│ • Danger: onslaught launch will burn all your cash
└─ YES → Are you competing on the same dimensions?
├─ YES → EXISTING MARKET
│ • Year 1 goal: market share
│ • Launch: onslaught (if you can afford Lanchester rules)
│ • Timeline: 12-18 months to profit
└─ NO → RESEGMENTED MARKET
• Low-cost or niche?
• Year 1 goal: share + education
• Launch: niche (focused onslaught)
Critical Numbers & Rules of Thumb
| Number | Rule |
|---|
| 9 of 10 | New product introductions that fail |
| 3-7 years | Time for a new market to generate profit |
| 12-18 months | Time for an existing market startup to generate cash |
| 3× | Spending needed to attack a monopolist (Lanchester) |
| 1.7× | Spending needed to attack weakest player in fragmented market |
| 74% | Market share threshold = effective monopoly |
| 41% | Market share threshold = guaranteed for new market creator |
| 26% | Below this, no player has real market influence |
| 47% | Market pioneer failure rate (Golder & Tellis) |
| 8% | Early market leader failure rate |
| ~13 years | Average late entry of early market leaders after pioneers |
| >50% | Startups where founding CEO is removed |
| ~9 months | Half-life of a startup VP of Sales post first customer ship |
| ≥2 cycles | Minimum iterations through Steps 1-2 for most startups |
Common Failure Patterns
| Pattern | Mechanism | Cure |
|---|
| "Build it and they will come" | No customer development process; assume product launch = customer acquisition | Customer Discovery before scaling |
| Premature scaling | Hiring full sales/marketing orgs before validating customers | Stay lean through Steps 1-2; "staff that fits in a phone booth" |
| Death spiral | Premature scaling → burn rate → missed numbers → fire VPs → fire CEO | Don't scale until you have a repeatable sales roadmap |
| Wrong Market Type | Using existing-market tactics in a new market (or vice versa) | Definitively determine Market Type; match every activity to it |
| Onslaught in new market | Massive launch spending for customers who don't exist yet | Early adopter launch; preserve cash; focus on adoption not share |
| Branding as strategy | Substituting "brand" for precise Customer Creation activities | Market Type-matched positioning, launch, and demand creation |
| Delegating to VP of Sales | Founders stop talking to customers | Founders in front of customers through at least first iteration |
| Feature-list driven product | MRD from focus groups instead of founder vision + customer validation | "First product for the few, not the many" |
| Kicking out founders | Process-oriented replacements gut innovation | Mission-centric organization; keep founders, add complementary execs |
| Confusing pipeline with roadmap | Measuring funnel metrics before knowing who buys or how | Build sales roadmap first; pipeline comes after |
The Webvan Cautionary Tale
$800M raised. Automated warehouses. $40M per distribution center. IPO at $8.5B market cap (larger than top 3 grocery chains combined). Bankrupt in 24 months.
What went wrong, step by step:
- No Customer Discovery - assumed customers wanted online grocery delivery; never validated
- No Customer Validation - committed to $1B Bechtel deal for 26 distribution centers before meaningful feedback
- Wrong Market Type - treated a new/resegmented market as an existing one
- Premature scaling - $18M proprietary software + $40M warehouse before shipping a single item
- Death spiral - 2,500 orders/day vs. 8,000 forecasted; 30% capacity utilization; $612.7M accumulated deficit
Counter-example: Tesco - used existing UK retail stores as launching pad, learned what customers wanted, found a profitable model. By 2002: 85,000 orders/week, $559M in sales. Fraction of Webvan's investment.
"Explicitly or implicitly, Tesco understood the process embodied by the Customer Development model."
Modern Relevance (2006 → 2026)
The framework's core ideas are durable. The specific examples and some assumptions are dated.
When Four Steps STILL Applies
- Any startup searching for customers and a business model
- Enterprise software with multi-stakeholder buying
- Hardware and physical products requiring customer education
- Deeptech / biotech with long development cycles (partial - Ch 1 explicitly excludes some biotech)
- Any product requiring behavior change in conservative industries
Where the Framework Shows Its Age
- PLG/bottoms-up SaaS - Customer Validation can happen via free tier metrics rather than founder-led sales calls. The iterative loop is faster.
- No internet-era distribution - written before app stores, cloud marketplaces, social distribution. Channel assumptions are dated.
- Lean Startup superseded the process - Eric Ries (one of Blank's first implementers at IMVU) formalized the Build-Measure-Learn loop with MVPs, making the iteration cycle faster and more explicit.
- AI-native products - mass consumer adoption can precede enterprise sales, breaking the earlyvangeist → mainstream sequence.
- Two-sided marketplaces - chicken-and-egg dynamics don't map cleanly to the four steps.
What Blank Got Permanently Right
- Customer Development as a parallel, recursive process alongside Product Development
- Market Type as the master variable that changes everything
- Premature scaling as the default killer of startups
- Founders must be in front of customers, not behind desks
- The pivot from Customer Validation back to Customer Discovery
- "No facts inside the building, only opinions"
Quick Reference Checklist
Customer Discovery:
Customer Validation:
Customer Creation:
Company Building:
The Big Idea
"Products developed with senior management out in front of customers early and often - win. Products handed off to a sales and marketing organization that has only been tangentially involved in the new product development process lose. It's that simple."
Customer Development is not a process for finding features. It's a process for finding customers, a market, and a viable business model - the three things that actually determine whether a startup lives or dies.
The four steps are iterative, not linear. Going backwards is learning, not failure. The whole process can be done by a team that fits in a phone booth. And no amount of funding can accelerate it.
"You cannot create a market or customer demand where there isn't any customer interest."
Supporting Files
- frameworks.md - Detailed Customer Discovery phases (0-4), six hypothesis templates, earlyvangelist identification, Customer Validation phases, sales roadmap construction, Market-Type questionnaire, New Lanchester rules, positioning by Market Type, launch selection, demand creation measurement
- cases.md - Webvan ($800M → bankruptcy), PhotosToYou (new-market branding disaster), FastOffice (death spiral through Plans B-D), InLook (premature VP Sales hiring), Design Within Reach ($180M success), Tesco (Webvan's antithesis), Furniture.com ($34M marketing vs $10.9M revenue), BetaSheet (founder/professional management tension)
- examples.md - Earlyvangelist pain hierarchy scoring, Market Type decision tree, sales roadmap template, Customer Validation pass/fail criteria, positioning templates by Market Type, launch type selection, Lanchester cost-of-entry calculator
- integration.md - Relationship to Crossing the Chasm (Four Steps covers what happens BEFORE the chasm), relationship to Lean Startup (Ries formalized Blank's process), conflicts with Mom Test (vision-first vs customer-first tension), conflicts with $100M Offers (offer-first vs discovery-first), Market Type vs TALC mapping
Honest Scope of the Book
- Published: 2005, Third Edition 2006
- Examples: Mostly late 1990s / early 2000s tech (Webvan, Palm, Handspring, PhotosToYou). Many companies no longer exist.
- Empirical base: Author's experience with 8 startups + VC advisory + teaching. Anecdotes and case studies, not statistical research.
- Where it shines: Pre-product/market-fit startups, enterprise B2B, any situation where "who are our customers?" is genuinely unknown.
- Where it's weak: PLG/freemium (iteration cycle is faster than the book assumes), consumer apps with viral distribution, two-sided marketplaces.
- Successor work: The Startup Owner's Manual (2012) is Blank's updated, more prescriptive version. The Lean Startup (Eric Ries, 2011) formalized many concepts with Build-Measure-Learn and MVPs.
- Author's context: Blank was a serial entrepreneur (8 startups including E.piphany), then became a professor at Stanford, Berkeley, and Columbia. The book emerged from his teaching.
The framework is foundational. The specific tactics and examples are dated. The core thesis - that customer and market discovery is a separate, iterative process that must run in parallel with product development - remains as true in 2026 as it was in 2005.
