growth-looplisted

# growth-loop Weekly GA-driven copy experiment loop for www.robertritacca.com (this repo's `website/` deploys there via Vercel). Each run: analyze last month's GA data, find ONE copy problem, form a hypothesis, implement the fix on a local branch, and write a clear report for the user to approve. **Never push, merge, or deploy — the user approves every change.** ## When invoked Run when asked to "run the growth loop" (`/growth-loop`) or by the `growth-loop-weekly` scheduled task. ## Scope guardrails (read first) - **Copy only.** Headlines, body text, CTA/link labels, button text, page `metadata` titles/descriptions — all inside `website/src`. No CSS, no layout, no component structure, no new components, no dependencies. - **One focused change per run.** One page, or one copy element (e.g. the same CTA wording) across a few pages. A reviewer should be able to read the diff in under two minutes. - **Local branch only.** Never `git push`, never merge, never touch the user's checked-out branch or working tree (use a worktree — see step 4). - Never read into version control or modify `ga-analysis/service-account.json` or `ga-analysis/output/`. ## The loop ### 0. Close the previous loop Read the newest report in `ga-analysis/loop-reports/` (git-ignored, local-only). If a previous experiment was approved/merged, check whether its metric moved in this run's data and record the verdict (improved / no change / worse / too early to tell) in this run's report. If the previous bra