How to build and run an effective growth team at a startup

Executive overview

Most early-stage companies are not ready for aggressive growth tactics. Without cohorts that show stable long-term retention, paid acquisition burns money without building a business.

Growth is not a team — it is a philosophy for running a company using data and experimentation. The core job is understanding the full funnel as a linked system, finding where leverage exists, and testing the right hypotheses rather than maximising test velocity.

The highest-leverage growth work is rarely new acquisition; it is fixing the quiet friction that stops already-interested users from reaching product value.

Readiness before growth

• Retention must show cohorts that genuinely level out before scaling acquisition
• Paid marketing on top of poor retention compounds the damage — expensive users who churn faster
• Build a simple spreadsheet model first: acquisition channels, activation rate, cohort retention, referral loops, monetization
• A 5% improvement in activation typically outweighs a 20% increase in top-of-funnel traffic
• Hire a smart analytical generalist who can hold the full business equation in their head — specialty skills are learnable, strategic intuition is not

Instrumentation and experimentation

• Instrument the site correctly before building an experiment framework; SQL-queryable data with integrity is the prerequisite
• Third-party A-B testing tools are now good enough that most teams should not build their own until they can justify a dedicated infrastructure engineer
• Start experiments at the user-account level — visitor-level or market-level experiments introduce statistical complexity that causes mistakes
• Speed of testing is not the primary goal; testing disparate hypotheses at the extremes of the experience yields larger effect sizes and faster statistical significance
• A backlog of 80 experiments often contains only 4–6 underlying hypotheses — validate or invalidate each hypothesis with 1–2 focused tests, then move on
• Send experiment results to the whole company with a write-up of why they won or lost; this builds experimentation culture more effectively than any mandate

Metrics and retention

• Start with a retention graph: does product value show up as a cohort curve that levels out?
• Add source tracking and an at-signup survey asking how users heard about the product — organic sources are often invisible to analytics
• After retention, find early proxies for retention so you can run activation experiments with fast feedback
• North Star metrics for marketplaces should combine both sides (e.g. transactions completed), not single-side engagement
• Measure liquidity at the market level: what percentage of requests receive a sufficient number of responses?
• NPS spikes sharply at the threshold where users feel the product delivered — instrument that threshold as a leading metric

High-intent conversion over homepage volume

• The homepage draws a mixed audience; most visitors are not in a booking or purchasing mindset
• High-intent flows (search results, referral sign-up, checkout) offer far better conversion leverage despite smaller scale
• Airbnb extended session cookie length from 2 days to unlimited, re-routing sensitive actions to re-authenticate — a hours-long code change that lifted revenue by over 1%
• Optimising signup and login is never "done"; for products where users arrive with a specific goal, friction at authentication is directly lost revenue

Growth channels and paid marketing

• At scale, roughly 70–80% of new customer acquisition comes from a single channel — build the product to optimise for that channel from the start
• Paid marketing is fast to launch and easy to test, but degrades over time: later cohorts are more expensive and retain less well
• Use paid as an accelerant (seeding new markets, rebalancing a marketplace) rather than as the core growth strategy
• SEO and word-of-mouth are more sustainable; Airbnb's word-of-mouth accounted for over 50% of guest acquisition and over 70% of host acquisition
• Viral growth through open platforms is largely saturated; referral programs (incentivised viral) still work but should be modelled like paid marketing with explicit payback periods
• In emerging markets, ad inventory on Facebook and Google is often underpriced because few premium brands have entered — a real arbitrage window for well-funded startups

B2B versus consumer growth

• B2B tactics increasingly mirror consumer ones, but with higher LTV allowing more creative acquisition spend
• If the total addressable customer list is in the hundreds of thousands or fewer, build the list and run every go-to-market channel against it directly
• SMB SaaS can scale through automated sign-up funnels and bottom-up adoption (e.g. Slack spreading through Airbnb organically)
• Enterprise requires identifying decision-makers first, using targeted content marketing and social proof to build credibility before a sales conversation

Team structure and organisational friction

• Most companies can operate longer without a VP of Marketing than they think
• Splitting retention ownership between a product growth team and a marketing team (owning email/push) creates disjointed user experiences — consolidate under product
• Acquisition is easier to keep separate because channels are distinct; retention and conversion should sit inside the product org
• A standalone growth SWAT team proving wins on one part of the product is often more effective than top-down mandates
• Publishing a weekly company-wide metrics email anchors all teams to the North Star and naturally surfaces funnel ownership

Ethics and long-term thinking

• Optimising for short-term engagement metrics (click-through, time on site) diverges from user interests; longer-term metrics re-align them
• Growth teams should have explicit principles: no dark patterns; always ask why a tactic is working before scaling it
• Products that charge users directly optimise for the user's outcome; ad-supported attention products face a structural misalignment
• The same growth tools being used for addictive engagement can be applied to health, fitness, and education products — the tactics are neutral