Custom Landing Page Generator

Meta Ad account ID: [YOUR AD ACCOUNT ID e.g. act_XXXXXXXXX]

You are going to build a custom landing page from my top-performing ads.

Follow these steps exactly.

STEP 1 — Pull top ads

Use GoMarble MCP to pull all ads from the last 30 days with spend above ₹500 (or $10 if USD account), sorted by spend descending, at the ad level.

Fields to pull: ad_id, ad_name, spend, purchase_roas, actions, campaign_name

STEP 2 — Apply volume floor

Remove any ad with fewer than 5 purchases (or 5 total conversions for lead gen accounts).

High ROAS on 2 purchases is noise, not signal. Only ads with 5+ conversions are real patterns.

STEP 3 — Score the remaining ads

Group surviving ads by product (use campaign name to identify the product group).

For each product group, calculate: Score = number of confirming ads × total conversions × average ROAS

Surface the top scoring pattern. That is the angle you will build around.

STEP 4 — Pull creative for the top 3 ads in the winning group

Use GoMarble MCP facebook_analyze_ad_creative_by_id_or_url for each ad.

Extract:

Exact hook / text hook (word for word)

Dominant emotion

Awareness level (cold / warm / solution-aware)

Core promise

CTA text

Landing page URL from the ad

STEP 5 — Auto-identify the product and pull product data

From the landing page URL extracted in Step 4, identify the store domain and product handle.

5a — Product data

First attempt the Shopify JSON endpoint: [store domain]/products/[handle].json

If it returns valid JSON, extract: title, all variant IDs, price, compare-at price, all image URLs, product description.

If it returns 404 or the store is not on Shopify, skip to 5b.

If no landing page URL was found in Step 4, pause and ask the user for the product page URL before continuing.

5b — Page scraping

If Apify is connected: use Apify RAG browser on the product page URL.

If Apify is not connected: use native curl to fetch the page directly.

Either way, extract: trust badges, guarantees, review count, COD availability, announcement bar copy, nav links, brand colors, fonts.

If the page is JavaScript-rendered and curl returns empty content, pause and flag this to the user. Apify will be needed to proceed.

STEP 6 — Recommend a template

Based on the winning angle type, recommend one of the following and explain why in two sentences:

Advertorial — parental concern / fear / problem-aware cold traffic Structure: authority opener → pain escalation → root cause reframe → product reveal → proof → offer

Short-form — solution-aware / benefit-led / warm traffic Structure: hero mirrors hook → product + bullets → proof → offer → CTA

Listicle — educational / warning signs / listicle hook Structure: "X signs" → problem list → product as answer → proof → CTA

Present your recommendation then ask:

"Should I proceed with [template name], or would you like a different approach?"

Wait for the user to confirm or suggest an alternative before moving to Step 7.

STEP 7 — Build the HTML page

Rules:

Hero headline = exact hook from the top-performing ad, word for word

CTA text = exact CTA from the top-performing ad

All product images pulled from real CDN URLs (no placeholders)

Real price and compare-at price from Shopify JSON

All CTAs link to: [store URL]/cart/[variant_id]:1

Match the brand's exact colors, fonts, and nav structure scraped from the live website

Include real announcement bar copy, real nav links, and real footer

COD mention minimum 4 times if this is an Indian DTC account

Nothing fabricated — use [INSERT REVIEW] placeholder if reviews are not available

STEP 8 — Output

Deliver a single complete HTML file. Confirm which template was used and why. State the winning angle and how many ads confirmed it.

Drop in your ad account ID and run.