diff --git a/skills/ecommerce-consultation.md b/skills/ecommerce-consultation.md
new file mode 100644
index 0000000..33ff17d
--- /dev/null
+++ b/skills/ecommerce-consultation.md
@@ -0,0 +1,95 @@
+---
+name: ecommerce-consultation
+description: Strategic ecommerce and conversion-rate-optimization consultation framework — how to diagnose a store, prioritize fixes by revenue impact, and turn findings into an actionable plan. Read when asked to advise on, audit, or grow a WooCommerce store rather than just execute a change.
+version: 1.0.0
+auto-load: false
+tags: [ecommerce, cro, strategy, consultation, revenue, conversion]
+---
+
+# Ecommerce Consultation — Diagnose, Prioritize, Act
+
+## When to use this skill
+
+Load when the ask is advisory, not mechanical: "why aren't we selling", "audit my store", "how do I grow this", "what should I fix first". For executing a specific change use `woocommerce` / `woodmart-mcp`. This skill turns data into a prioritized plan.
+
+---
+
+## Principle: every recommendation ties to revenue
+
+Revenue = Traffic × Conversion Rate × Average Order Value × Purchase Frequency.
+
+Every finding must name which lever it moves and roughly how much. "Improve the product page" is not a recommendation; "the product page has no reviews block, and reviews lift conversion on considered purchases — adding it targets the CR lever" is. Lead with the lever, quantify when you can, and always end with a concrete next action.
+
+---
+
+## Step 1 — pull the real numbers first
+
+Never advise from assumption. Ground the audit in live data:
+
+- `wc_sales_report` — revenue, orders, items sold, AOV (gross ÷ orders) over a meaningful window.
+- `wc_get_store_stats` — broader snapshot.
+- `wc_list_products` — catalog size, what's published, pricing spread.
+- `wc_list_orders` — recent order flow, statuses, abandoned/failed patterns.
+- `wc_list_product_reviews` — social proof depth.
+- `wc_list_payment_gateways`, `wc_list_shipping_zones`, `wc_list_tax_rates` — checkout friction sources.
+- For traffic/behavior, pull from the connected analytics (GA4) rather than guessing demand.
+
+If the store has near-zero orders, say so plainly and shift the diagnosis to "is anyone arriving, and can they check out" before touching optimization.
+
+---
+
+## Step 2 — diagnose by funnel stage
+
+Walk the funnel; the biggest leak is the priority, regardless of where it sits.
+
+1. **Arrival** — is there qualified traffic? No traffic is an acquisition problem, not a CRO one. Check GA4 sources.
+2. **Catalog / discovery** — can visitors find and trust products? Category structure (`wc_list_product_categories`), search, filters (Woodmart shop layout), imagery quality.
+3. **Product page** — the conversion crux: real photos, clear price, benefit-led copy, reviews, trust signals, obvious add-to-cart.
+4. **Cart** — surprise costs kill carts. Shipping and tax surfaced early? Any upsell/cross-sell?
+5. **Checkout** — gateways enabled and ordered sensibly, shipping zone covers the customer's region, minimal fields, guest checkout allowed.
+6. **Retention** — post-purchase email, repeat-purchase incentives, win-back. Frequency is the cheapest growth lever once the rest works.
+
+---
+
+## Step 3 — prioritize by impact × effort
+
+Rank fixes, don't dump a list. For each finding score impact (revenue lever moved) against effort (tool change vs theme change vs new build). Order:
+
+1. **High impact / low effort** — do now. (A disabled gateway, a missing reviews block, an uncapped bug.)
+2. **High impact / high effort** — plan and schedule. (Rebuilding the product-page template, a new acquisition channel.)
+3. **Low impact / low effort** — batch into a cleanup pass.
+4. **Low impact / high effort** — drop.
+
+Present the top 3–5 only. A 20-item audit no one acts on is worth less than three fixes shipped.
+
+---
+
+## Step 4 — turn each finding into an action
+
+Write every recommendation so it can be executed without a follow-up question. Name the lever, the fix, the tool, and the expected direction:
+
+> **Checkout (CR lever):** Only one payment gateway is enabled and it's a manual method, forcing bank transfer. Enable a card gateway and set it as default (`wc_list_payment_gateways` to confirm, then enable in store settings). Expect a direct lift in completed checkouts.
+
+That is a plan. "Improve checkout" is not.
+
+---
+
+## Marketing output rule
+
+When the consultation produces any ad, caption, email, or promo copy, write it conversion-first and always include a clear call to action — never leave copy without the next step for the buyer. Keep it high-impact and concise; lead with the offer, close with the action. (This matches the account's standing preference for a strong CTA on all marketing output.)
+
+---
+
+## Deliverable shape
+
+A consultation output is: a one-line verdict (what's really going on), the live numbers it's based on, the top 3–5 prioritized fixes each with lever + action + tool, and the single thing to do first. End with that first action, stated plainly. Record durable findings to the store's memory (`memory_set`) so the next session doesn't re-audit from scratch.
+
+---
+
+## Anti-patterns
+
+- **Advising without data** — pull `wc_sales_report` and GA4 first, always.
+- **A flat 20-item list** — rank by impact; surface the top few.
+- **CRO on a no-traffic store** — fix arrival and checkout-capability first.
+- **Vague recommendations** — every one names the tool and the expected result.
+- **Optimizing AOV/frequency before the funnel converts at all** — sequence matters; fix the biggest leak first.