funnel-analysislisted

# Funnel Analysis Before writing your answer, read `examples/output.md` in this skill directory and match its density, structure and tone. It is the reference for what a good run of this skill looks like. **Before anything else: emit no text until the report.** **Your first action is a tool call, not a sentence** — not "State directory is empty, running discovery", not "Let me start with the overview", not "Now writing state files, then the report". Writing state before the report is something you do, not something you announce. And nothing between calls either: no "Drop confirmed, moving to channels", no "Drilling into campaigns", no "Checking seasonality". The user reads every one of those before your answer, and a run that narrates its way to a conclusion reads as one that has not reached it. Make the calls in silence; your first and only message is the finished report. **And nothing after it:** write the profile, the ledger and the run log *before* the report, never once it is written. A tool call after the report forces a second message, and a run that logged its diagnosis first and then added "Diagnosis complete: the drop traces to /collections/sale" made the user read the same finding twice. Find the weakest funnel stage and isolate its cause. Budget: ≤10 calls. Vertical playbooks: `skills/seal-copilot/references/ecommerce-playbook.md` (stores) and `skills/seal-copilot/references/hotels-playbook.md` (hotels). **Resolve the site before any call that takes a `site_id