a builder's codex
codex · operators · Elena Verna · ins_add-new-growth-model-every-18-months

Add a new growth model every 18 months and protect it from KPIs for 12

By Elena Verna · Growth advisor; former Dropbox, Miro, Amplitude, SurveyMonkey · 2026-04-28 · podcast · Elena Verna 3.0 — 10 growth tactics that never work — Lenny's Podcast

Tier A · TL;DR
Add a new growth model every 18 months and protect it from KPIs for 12

Claim

Every loop decays. Andrew Chen's Law of Shitty Click-throughs — over-optimizing a single channel produces diminishing returns — applies to growth loops as well. Allocate 20–25% of the growth team's annual time to introducing new loops with no metric expectation for the first 12–18 months. Without a runway, every new loop dies before it has a chance to compound.

Mechanism

A loop's economics are non-linear in time. Months 1–12 produce noisy data and small results. Months 18–36 are where compound effects appear. Most teams kill loops at month 6 because they aren't pacing the existing loops. Pre-allocating capacity and removing the ROI gate during the seeding period preserves enough loops to ride out their slow start.

Conditions

Holds when:

Fails when:

Evidence

"Most loops decay over 5–7 years; some, like Dropbox sharing, don't. Allocate 20–25% of the growth team's annual time to introducing new loops with no metric expectation for the first 12–18 months."

Elena cites her own experience across Dropbox, Miro, SurveyMonkey, and Amplitude — the durable loops are always the ones that got the multi-year runway, not the ones launched and judged in a quarter.

— Elena Verna on Lenny's Podcast, 2026-04-28

Signals

Counter-evidence

For pre-PMF or distressed companies, this rule is a luxury. Elena explicitly notes "growth teams cannot fix declining businesses." Add new growth models only when the existing ones are healthy enough to fund the experimentation.

Cross-references

Open the interactive view → View original source → Markdown source →