Growth for startups: product-market fit, channels, and decisions

Executive overview

Most startups don't have product-market fit, and working on growth before you do is a waste. The single most reliable signal of PMF is retention — not surveys, NPS, or registered users.

Once retention holds, growth comes from one or two channels, not all of them. Decisions at scale require data and experimentation, not the loudest voice in the room.

The founders who win are the ones willing to do things that don't scale first.

Do things that don't scale

• Early Airbnb founders flew to New York, photographed listings themselves, and interviewed hosts in person.
• Direct customer contact surfaces bugs, UX failures, and trust gaps that no dashboard reveals.
• The goal: make the product work for a small group of real users, who then become the foundation for growth.
• Most companies begin with friends, then friends of friends — not Google Ads.
• Unlearn the MBA instinct that only scalable things are worth doing.

Measuring product-market fit with retention

• Identify the metric that represents the core value of your product (bookings, rides, payroll runs, daily active views).
• Identify the natural frequency of that behaviour (daily, weekly, monthly, annually).
• Plot a cohort retention curve: what share of week-0 users return in week 1, week 2, week N?
• A bad product curves to zero. A good product flattens — meaning a stable share keeps returning.
• DoorDash: 20% monthly retention after 18 months. GitHub: 30% after 60 months. Both flat lines.

Metrics that lie about PMF

• Registered users — says nothing about repeat usage.
• Visitors — says nothing about value delivered.
• Conversion rate — says nothing about who you're converting.
• Free usage of a product that should be paid — users who won't pay won't sustain you.
• NPS — correlates with brand perception, not product retention. The iPhone has a bad NPS score.
• Surveys — useful for product improvement, not for PMF measurement.

Conversion rate optimisation

• Every step of your product funnel has a drop-off rate. Measure it.
• Airbnb mapped four pages (P1 home → P2 search → P3 listing → P4 booking) and worked to move users through them.
• Common drop-off causes: content mismatch (wrong audience), browser/device bugs, friction in sign-up flow.
• Key areas to work on: internationalisation, authentication, onboarding, purchase conversion.
• Copy the best-in-class flows (Pinterest, Airbnb) — those teams have already optimised what you're building.

Growth channels

• Most large companies grew through one or two channels, not all of them. TripAdvisor = SEO. Pinterest = SEO. Airbnb = referrals + virality.
• Pick the channel that matches your product type, then go deep.

Rare-behaviour products → search (SEO/SEM)

• If users only need your product occasionally, they'll Google it.
• Get on Google through paid search (SEM) or organic ranking (SEO).

Naturally shareable products → virality and referrals

• Some products are exciting to talk about (Lyft, Airbnb). Lean into that.
• Referrals = financial incentive to invite friends. Airbnb's: $40 to the new user, $20 to the referrer.
• Measure every step of the referral funnel: offer shown → page visited → invite sent → invite clicked → signed up → booked.
• Optimise the invite email: sender name (friend, not brand), clear value prop above the fold, deadline for urgency, exclusive CTA ("accept my invitation"), social proof.

Social / network-effect products → viral loops

• More users = better product (LinkedIn, Facebook).
• At sign-up, prompt users to invite contacts immediately — their experience depends on it.

Small, known addressable market → sales

• If you can list all your customers, make that list. Names, decision-makers, emails, phone numbers.
• Start contacting them directly. Don't broadcast to the world when only a few hundred people are relevant.

High LTV products → paid online marketing

• Only spend on ads if you have paying customers. The most common mistake is buying ads before you have revenue.
• Key metrics: CAC (customer acquisition cost) and LTV (lifetime value). LTV must exceed CAC.
• Don't wait 8 months to validate. Set a short payback window (1–3 months) as your early target.
• Main channels: Google, Facebook, Instagram.

Search engine optimisation

• SEO is a zero-sum game — you're competing against incumbents (TripAdvisor, Pinterest) who rank for everything.
• New keywords are the exception: emerging topics have no established rankers yet.
• Two levers: on-page (title tags, crawlability, keyword targeting, error-free pages) and off-page (domain authority, inbound links from high-authority sources).
• Start with keyword research. Find what people actually search for before writing content.
• Inbound links from press and credible sites raise your domain authority and lift all rankings.

A/B testing and data-driven decisions

• A/B testing is irrelevant until you have meaningful traffic. Check with an online calculator first — most early-stage startups won't meet the threshold.
• The point: create two parallel universes (control vs. experiment) and measure which performs better.
• Avoids arguments about what caused a metric change. Removes gut-feel dominance.
• Airbnb example 1: custom share sheet vs. native iOS share sheet. Custom version drove 40% more shares — the opposite of what most people predicted.
• Airbnb example 2: no sign-up wall vs. dismissible wall vs. hard wall. Hard sign-up wall produced 2.6% more iOS bookings — the least popular prediction.
• Build a culture of experimentation: data decides, not the loudest voice.