Metrics and go-to-market
Owned audiences
Email, docs, community and open source compound. Rented channels can be turned off overnight. Engineering-as-marketing is a channel too.
Rented channels (ads, social algorithms, app stores) can be turned down by someone else overnight. Owned assets compound: email and SMS lists, content and docs that rank, community where trust lives, open-source repos you steward, and brand that generates direct traffic.
Owned audiences lower CAC over time and fund growth loops. They are slow — which is why startups still buy paid spikes — but they are what make a business defensible when CPCs double. See aggregation theory: if Platform X is your only path, assume they will tax you or clone you.
Concept
An owned audience is a relationship you can reach without bidding or begging a platform's algorithm: permission (email, notifications), artifacts you host (docs, tools), or a community you steward. "Community-led growth" without a path to activation and revenue is a fan club.
Engineering as marketing
Weinberg's catalogue includes engineering as marketing: free tools, calculators, open datasets, status pages, lint rules, or utilities that attract the ICP and capture emails. Stripe's early developer documentation and Notion's templates follow the pattern — the artifact is useful even before the paid product, and it ranks or gets shared.
Product-led channels deserve the same CAC scrutiny as ads:
- Invites / collaboration loops — measure invites sent, invite→signup, signup→activation.
- Marketplace listings — include rev-share and co-sell time in CAC.
- Content that ranks — attribute trial starts from organic landing pages; ignore vanity traffic that never activates.
Walking through a free tool
You ship a free "LLM token cost estimator." It drives 20,000 visits/month. 4% start a trial (800), 12% of trials pay ($50/mo SMB plan) → 96 customers/month → ~$57.6k ARR added per month from this asset.
Fully loaded eng+design cost to build and maintain was $40k once plus $2k/month. First-year fully loaded cost ≈ $40k + $24k = $64k for roughly $57.6k × 12 ≈ $691k ARR if retention holds — spectacular CAC — if those customers match ICP and renew. If they are students who churn in 60 days, the channel fails GRR even while "signups" look viral.
Engineers own more of distribution than they admit: docs quality (SEO), referral deep links, invite UX, marketplace listings, instrumentation for funnel drop-off, open-source DX. A public template gallery or a meaningful integration can be higher-leverage "marketing" than another homepage rewrite.
Common mistakes
- Vanity member counts. Community without activation is a fan club.
- Vanity traffic. Rankings that never trial are not a channel.
- Student signups that churn in 60 days. Spectacular top-of-funnel, failed GRR.
Check yourself
20,000 visits, 4% trial, 12% of trials pay at $50/month. Monthly ARR added?
About $57.6k. 20,000 × 0.04 × 0.12 × $50 = $4,800 MRR ≈ $57.6k ARR.
Why is a ranking page not automatically a good channel?
If visitors never activate or they churn immediately, you bought vanity traffic, not customers.
Key takeaways
- Owned audiences (lists, docs, community, OSS, brand) compound and cannot be switched off overnight.
- Engineering-as-marketing is a real channel. Treat it with CAC math.
- Tie community and content to activation and revenue, not vanity counts.
- Spectacular signups that churn fail GRR. Quality of the customer matters.
Further learning
Andrew Chenandrewchenandrewchen.com Y Combinator LibraryYC's library of startup advice: essays and videos that teach you how to start a company.YC Startup Library