Digital Product Sales Page Offer Extractor

Turn one public product page or pasted sales copy into structured offer JSON with exact source evidence. Use it to audit Gumroad, Payhip, Lemon Squeezy, course, template, starter-kit, prompt-pack, and boilerplate offers without inventing unstated claims.

What does this digital product offer extractor do?

The Actor reads the buyer-visible text from one public HTTP/HTTPS page or from pasted copy. It extracts the product name, target buyer, promise, deliverables, visible prices, purchase actions, delivery details, requirements, refund terms, and support paths.

Every extracted value includes confidence and exact supporting evidence. When the page does not state a fact, the value stays empty and the field appears in missingFields. This makes the output useful for quality assurance, structured catalogs, comparison workflows, and agent tools that must distinguish source facts from assumptions.

Why use it?

• Audit before launch: find missing buyer-facing facts before sending traffic to a sales page.
• Normalize product catalogs: convert inconsistent creator listings into a stable JSON shape.
• Ground an AI workflow: give an API, webhook, or MCP client evidence-backed offer facts instead of generated summaries.
• Review client pages: let an agency compare what a seller intended to communicate with what the page actually states.
• Monitor offer changes: schedule repeated runs and compare structured outputs in your own workflow.

This is not a generic retail scraper and it does not rewrite the page. It focuses on the offer a buyer can verify.

Input

Provide either:

1. one public HTTP/HTTPS product-page URL; or
2. up to 200,000 characters of buyer-visible pasted text.

Pasted text takes precedence when both are supplied. The included Store example uses pasted first-party sample copy, so a first run does not depend on a third-party website.

{
  "text": "Creator Workflow Kit for solo consultants. Includes 12 checklists and three client handoff templates. Launch price: $19. Download immediately after purchase. Email support is included."
}

Output

One successful run returns one dataset record containing:

• product name and target buyer;
• promise and deliverables;
• visible prices and purchase actions;
• delivery details and requirements;
• refund terms and support paths;
• confidence and exact evidence for each extracted value;
• a missingFields list for facts the page does not state.

Example shape:

{
  "productName": {
    "value": "Creator Workflow Kit",
    "evidence": "Creator Workflow Kit for solo consultants."
  },
  "prices": [
    {
      "value": "$19",
      "evidence": "Launch price: $19."
    }
  ],
  "missingFields": ["refundTerms"]
}

Missing values remain empty. The Actor does not invent sales claims or treat a generic navigation label as a purchase action.

Run it as an API or automation

Use the Apify Console, API, scheduled Tasks, webhooks, Google Sheets integrations, Make, n8n, or an Apify MCP connection. The input and dataset schemas keep the same top-level contract for URL and pasted-text modes.

A successful dataset delivery emits one pay-per-event offer-extracted event. Failed, rejected, or empty extractions deliver no paid result. Review current pricing on the Actor's Pricing tab before running it.

Common workflows

Audit a Gumroad or Payhip page

Submit the public listing URL. Review missingFields first, then inspect the exact evidence for prices, deliverables, purchase actions, delivery, refunds, and support.

Check a draft before publishing

Paste the draft sales copy. This avoids exposing a private preview URL and shows which buyer-facing facts are still absent.

Feed a catalog or agent

Send the dataset record to your database, webhook, spreadsheet, or MCP workflow. Use the evidence fields to preserve traceability.

Limits and safety

• One public page or pasted text per run
• No login, CAPTCHA, paywall, or access-control bypass
• No private, local, or credential-bearing URLs
• No proxy rotation
• No legal, compliance, or claim-verification advice
• Not affiliated with the page, seller, marketplace, or platform being analyzed

If a public page blocks automated access, use pasted buyer-visible copy that you are authorized to process.

Troubleshooting

The result is empty: confirm the page contains buyer-visible offer text, or use pasted text.

A field is missing: the page may not state it. Check missingFields and the evidence entries before assuming an extraction error.

The URL is rejected: only public HTTP/HTTPS pages are accepted. Local, private-network, credential-bearing, and non-HTML targets are rejected by design.

The website changes: rerun the Actor and compare the new dataset record in your own storage or monitoring workflow.