How to find product-market fit and build a growth engine with Sean Ellis

Executive overview

Most products fail because no one actually needs them. Sean Ellis's survey question — "How would you feel if you could no longer use this product?" — gives founders an early read on whether they've built something people consider a must-have, before retention data has had time to accumulate.

Hitting 40% "very disappointed" is a signal to start thinking about growth, not a green light to scale. The real work is understanding who those users are, what benefit they value, and why that benefit matters — then rebuilding onboarding, positioning, and acquisition around that insight.

The biggest growth lever is usually not acquisition — it's collapsing the time between signup and the first experience of core value.

The Sean Ellis test: origin and mechanics

• The question flips a satisfaction scale: "very disappointed" users self-select as the people who genuinely need the product
• 40% is a pattern-derived threshold, not a scientific law — adjust for cultural context (Nubank uses 50% for Brazilian users)
• It is a leading indicator; actual retention cohorts are the lagging confirmation
• Survey only users who have activated (used the product at least twice, within the past two weeks)
• Sample size minimum: ~30 responses before drawing conclusions
• One-off products (single events, one-time experiences) are not suited to this question; use NPS as a filter instead

How to use the score — at any percentage

• Whether the score is 7% or 70%, the most valuable step is always the same: drill into the "very disappointed" cohort
• Ask open-ended: "What is the primary benefit you get?" — crowdsource to find recurring themes
• Run a follow-up survey with multiple-choice benefit statements; ask "Why is that benefit important to you?"
• The why surfaces the emotional context that powers great acquisition copy (e.g. "drowning in email" for a search product)
• Ignore "somewhat disappointed" users when shaping the core product — their feedback risks diluting the must-have experience
• Superhuman's refinement: find "somewhat disappointed" users who share the must-have benefit and ask what's missing — this lifts fence-sitters without compromising the core

Moving from low scores to 40%+

• Lookout case: repositioned from a multi-feature security app to an antivirus product in two weeks — score jumped from 7% to 40%
• Two levers: (1) set the right expectations upfront so only the right users convert, (2) streamline onboarding so users reach the core value immediately
• Repositioning alone, without changing the product, can dramatically shift the score
• Speed to value is almost always the highest-leverage onboarding variable

Sequencing growth after product-market fit

• Customer acquisition is the last thing to scale, not the first
• Recommended sequence: activation → engagement loops → referral → revenue model → acquisition at scale
• Rationale: if you can't convert, retain, and monetize, paid acquisition will always be too expensive to scale profitably

Activation

• LogMeIn: 95% of signups never completed a remote session; freezing the product roadmap to fix onboarding improved signup-to-usage 10x in three months; the same channels that capped at $10k/month then scaled to $1M/month
• Define your activation metric qualitatively first (what experience proves the product is working?), then validate with retention correlation
• Avoid activation metrics set too far downstream — they are accurate but not actionable
• Two levers on any activation funnel: increase desire, reduce friction
• When A/B tests stall, ask drop-off users directly: "What happened?" — one such question at LogMeIn revealed a credibility problem, leading to a 300% improvement in download rate

North Star metric

• Derive it from the must-have benefit uncovered by the survey — it should reflect units of value delivered, not revenue
• It must be directional (up and to the right), not a ratio
• Revenue should correlate with the North Star, not be it
• Examples: Amazon — monthly purchases; Airbnb — nights booked; Uber — weekly rides; Facebook shift from MAU to DAU changed product behavior significantly
• A 30-minute team exercise with the right framework is sufficient; don't let this drag on

Growth channels and loops

• Pick channels based on where your must-have users actually go to find products like yours — ask them directly
• Demand generation (interruption, in-context) vs. demand harvesting (search) are distinct strategies requiring different execution
• Referral programs accelerate existing word-of-mouth; they cannot create it where it doesn't exist
• Freemium works only if the free product is genuinely great — a weak free tier produces weak word-of-mouth
• Dropbox referral: success depended on strong organic sharing that already existed pre-program

How growth has changed

• In the early days, being data-driven on acquisition alone was enough to win
• Today, most teams are data-driven; the edge comes from cross-functional testing across the full growth engine — activation, engagement, referral, revenue, and acquisition
• The hardest part is organisational: product and marketing teams are not accustomed to collaborating on shared experiments
• Companies that built this cross-functional muscle early maintain a lasting structural advantage

ICE prioritisation

• ICE (Impact, Confidence, Ease) was designed to let anyone across the company submit experiment ideas and understand why an idea was or wasn't chosen
• "Reach" in RICE is already captured within Impact — adding it risks over-engineering a framework whose value is its simplicity
• Better experiments come from better questions, not more prioritisation complexity
• AI will increasingly help model outcome probabilities and reduce the bottleneck at the analysis stage