Claim
Algorithm channels (paid search, paid social, SEO) are rented attention; the cost trends up forever and the rent goes to the platform. Earned channels — product-led virality, sharing loops, user-generated content, referral mechanics — accrue to you. With AI eating the search UI, algorithm-channel costs will rise faster; the time to invest in earned channels was yesterday.
Mechanism
Algorithm channels are auctions. As more bidders enter, CPM rises; the only escape is having the highest LTV in the auction, which compresses margins. Earned channels are loops you own: each transaction or share is an asset that produces future transactions or shares without re-paying the platform. Dropbox's sharing loop has driven 50% of acquisition for 17 years; the cost per acquisition there is fixed engineering work, not rising auction prices.
Conditions
Holds when:
- The product has a natural sharing or content-generation moment.
- The team can invest 12–18 months in a new loop without expecting metric impact.
Fails when:
- The product is genuinely B2B-stealth (compliance, security) — sharing loops don't fit.
- The team is pre-PMF; loops amplify weak product faster than algorithm channels do.
Evidence
"When you're doing organic search or paid search, you're making Google richer. With AI eating the search UI, your cost of acquisition is only going to go up. You're constantly going to be praying to algorithm gods."
Dropbox's sharing loop = 50% of acquisition for 17 years. Most loops decay in 5–7 years; the durable ones don't.
— Elena Verna on Lenny's Podcast, 2026-04-28
Signals
- Earned-channel acquisition share grows quarter over quarter.
- The team has at least one loop more than 3 years old still producing.
- Algorithm-channel CAC is treated as a rising cost to plan around, not a constant.
Counter-evidence
Some categories are stuck in algorithm channels because the buyer journey is search-shaped (legal, accounting, dental services). Earned-channel investment there underperforms; allocate accordingly. The rule applies to products with natural sharing surfaces, not all products.
Cross-references
- Add a new growth model every 18 months and protect it from KPIs for 12 — the cadence implication
- Don't hire a head of growth before PMF or to fix a declining business — the prerequisite (PMF before earned-channel work)