REFINE Full-Funnel Dashboard

Objective. Produce the populated Full-Funnel Dashboard: every step's primary metric run down one column, read against the benchmark band its chapter settled on, with the weakest lever in the chain flagged as this cycle's candidate constraint.

Inputs this skill needs

• [Playbook Assets] — the company's populated playbook across all nine steps. The dashboard consolidates the benchmark each step chapter set; where the company's live figures are not supplied, leave the metric cell as ___ for the reader to fill from their analytics platform.
• No Foundation slots are read directly. This is a Refine-wave skill consolidating the framework's own benchmarks.

ROCKET prompt

ROLE: You are a growth analyst building the operational heart of the REFINE loop — the Full-Funnel Dashboard the chapter describes as "the consolidation of every benchmark this book has given you, gathered into one view." You are the ELEVATE Scorecard made concrete and personal.

OBJECTIVE: Produce the Full-Funnel Dashboard as a single table — one row per lever, the company's metric run down the Your metric column (or ___ where unknown), each read against the same benchmark band its step chapter taught, with a one-line diagnosis of a weak number and the source named in-chapter — then flag the candidate constraint.

CONTEXT: The chapter is explicit that these are not new numbers — they are "the scattered benchmarks of the whole book brought together so you can see the complete chain at once." Reproduce the exact bands the dashboard in book/5-refine/5-refine.md carries, row for row, including the dual SELL rows (conversion + checkout completion), the dual NURTURE rows (sequence engagement + cart recovery), the dual UPSELL rows (take-rate + AOV uplift), the dual EDUCATE rows (90-day repeat + NPS/CSAT) and the dual SHARE rows (referred-customer LTV + reviews→conversion). Read in two passes as the chapter teaches: the absolute pass (in band / below / above) then the comparative pass (which gap is largest and earliest). Honour the warning that a figure far above a band — a GIFT opt-in well over range — can mean freebie-seekers, not a win. Carry every [VERIFY] flag the chapter's bands carry; benchmarks drift.

KEY INSTRUCTIONS:

1. Build the dashboard table with these columns exactly: | Step | Lever governed | Your metric | Healthy benchmark band | One-line diagnosis of a weak number | Source (as settled in-chapter) |.
2. Include every row the chapter's dashboard lists, in funnel order: HOOK · GIFT · IDENTIFY · ENGAGE · SELL (conversion) · SELL (checkout completion) · NURTURE (sequence engagement) · NURTURE (cart recovery) · UPSELL (take-rate) · UPSELL (AOV uplift) · EDUCATE (90-day repeat) · EDUCATE (NPS/CSAT) · SHARE (referred-customer LTV) · SHARE (reviews→conversion).
3. Reproduce each benchmark band exactly as the chapter states it (e.g. paid-social CTR ~0.9–1.8%; GIFT opt-in ~20–40% good / ~10% average; cart abandonment ~68–72%; SELL conversion ~2–3%, strong ~5%+; abandoned-cart recovery ~10%; UPSELL take-rate ~10–30%; 90-day repeat ~25–40%). Keep ranges; never invent precision.
4. Where the company's live figure is supplied via [Playbook Assets], place it in Your metric; otherwise write ___ and note in the intro that the reader must populate from their analytics platform with current-period figures, not estimates.
5. After the table, write the two-pass read: (a) absolute — mark each row in band / below / above; (b) comparative — name the row with the largest gap below its band that sits earliest in the chain. That row is this cycle's candidate constraint.
6. Flag any figure suspiciously far above its band as a possible lead-quality warning, not a win, per the chapter.
7. Carry the chapter's standing caveat verbatim in spirit: bands are orientation not targets, every figure inherits its [VERIFY] flag, date-stamp any internal goal, refresh quarterly.
8. No fabricated proof — cite only the sources the chapter names (WordStream/Meta, Mailchimp, Unbounce, Klaviyo, Baymard, Bain/Reichheld, Spiegel, etc.) with their years and [VERIFY] flags intact.

EXAMPLES (generic, illustrative shapes only):

• Row: | GIFT | Landing-page opt-in rate | ___% | ~20–40% matched page; ~10% average; below ~10% underperforming | Promise not specific; hook–gift mismatch; no credible quick win | Unbounce / HubSpot benchmarks [VERIFY] |
• Read: "SELL conversion sits at ___ vs ~2–3%; GIFT opt-in sits below ~10%. The earliest, largest gap is GIFT — that is the candidate constraint; confirm via the cascade before acting."

TONE & FORMAT: Analytical, exact, diagnostic; British English; defer to elevate-voice for prose. Output the structure defined in the Output contract.

Output contract

Write one Markdown file to companies/<slug>/playbook/refine/metrics-dashboard.md with this exact shape:

• # REFINE Full-Funnel Dashboard (H1)
• A short intro paragraph: what the dashboard consolidates, and the instruction to populate ___ cells with current-period figures from the reader's analytics platform.
• ## The dashboard — the consolidation table with the six columns named above and all fourteen rows in funnel order, each band reproduced from the chapter with [VERIFY] flags intact.
• ## How to read it — the two-pass read: an absolute pass (in band / below / above) and a comparative pass naming the candidate constraint (largest gap, earliest in chain).
• ## Candidate constraint — one short paragraph naming the flagged constraint row and the next action (run the metrics cascade to confirm cause before fixing).
• ## Caveats — 3 bullets: bands are orientation not targets; verify each figure against its current source before public citation; refresh quarterly and date-stamp internal goals.

Total length under 900 words. Conforms to _shared/asset-schema.md (returned as the markdown field).