CSRD & ESG Data Extractor API (Scope 1, 2, 3)

Automate the extraction of sustainability metrics from complex Annual Reports (PDF & HTML). Designed for ESG Consultants, Carbon Accounting Platforms, and Financial Analysts.

---

🚀 Why use this Actor?

With the CSRD (Corporate Sustainability Reporting Directive) deadline approaching, extracting data manually from 200+ page PDF reports is slow, expensive, and error-prone.

This Actor is an enterprise-grade extraction engine that navigates corporate websites, downloads Sustainability/Annual Reports, and uses AI to extract structured Scope 1, 2, and 3 emissions data with high precision.

Key Features

• 📄 Advanced PDF Parsing: Unlike simple HTML scrapers, this actor downloads and processes heavy PDF files (OCR capabilities included for scanned tables).
• 🛡️ Anti-Blocking Technology: Built on top of Puppeteer with Residential Proxies and stealth plugins to bypass Cloudflare and strict corporate firewalls.
• 🎯 Scope 1, 2, 3 Granularity: Extracts specific emission figures, units (tCO2e), and reporting years.
• 🔍 Audit Trail & Citations: Every extracted data point includes the source context (text snippet or page reference) so you can verify the numbers for compliance.
• 📊 Reliability Scoring: Returns a confidence score and reasoning for every extraction, flagging potential data gaps.

---

🛠️ How it works

1. Input: You provide the domain (e.g., volvocars.com) and the company_name.
2. Discovery: The actor scans the website to find the latest "Sustainability Report", "Non-Financial Statement", or "Annual Report".
3. Processing: It downloads the document (handling PDF or HTML).
4. Extraction: Using LLM-powered analysis, it identifies ESG tables and relevant paragraphs.
5. Output: You receive a clean JSON with the structured data.

---

📥 Input Parameters

The input of this actor should be JSON.

Field	Type	Description
domain	String	Required. The website of the company (e.g., volvocars.com).
company_name	String	Required. The full name of the company to aid the search.
reporting_year	String	Optional. Specific year to target (e.g., 2023). Defaults to the latest available.
force_pdf_processing	Boolean	Optional. If true, prioritizes PDF documents over HTML pages. Default: true.

Example Input

{
    "domain": "volvocars.com",
    "company_name": "Volvo Car Corporation",
    "reporting_year": "2023",
    "force_scrape": true
}

---

📤 Output Example

The results are stored in the default dataset associated with the run. Note how the actor distinguishes between market-based Scope 2 and the massive Scope 3 categories typical of automotive companies.

[
  {
    "domain": "volvocars.com",
    "company_name": "Volvo Car Corporation",
    "reporting_year": 2023,
    "status": "success",
    "data": {
      "emissions": {
        "scope_1": {
          "value": 38000,
          "unit": "tCO2e",
          "context": "Direct GHG emissions from manufacturing and operations (Page 182, Sustainability Notes)",
          "confidence": "High"
        },
        "scope_2": {
          "value": 12000,
          "type": "market-based",
          "unit": "tCO2e",
          "context": "Indirect emissions from purchased electricity, heating and cooling (market-based). Location-based was 85,000 tCO2e.",
          "confidence": "High"
        },
        "scope_3": {
          "value": 42500000,
          "unit": "tCO2e",
          "categories_included": ["Purchased goods and services", "Use of sold products", "Upstream transportation"],
          "confidence": "High",
          "notes": "Includes lifecycle emissions from sold vehicles."
        }
      },
      "reliability_score": 0.98,
      "reliability_reasoning": "Data extracted explicitly from the 'GRI Content Index' and 'Greenhouse Gas Emissions' tables in the Annual Report 2023."
    },
    "source_url": "https://investors.volvocars.com/annual-report-2023.pdf",
    "scraped_at": "2024-05-20T14:30:00Z"
  }
]

---

💰 Pricing & Cost Efficiency

This actor is designed to be significantly cheaper than manual data entry.

• Manual Entry: An analyst takes ~30-60 minutes to find and transcribe Scope 1-3 data per report. Cost: ~$50/report (labor).
• Apify Actor: Takes ~1-3 minutes. Cost: Fraction of manual labor.

Recommended for bulk usage. If you need to process 100+ companies, please contact me via the Issues tab for a custom solution.

---

⚠️ Known Limitations

• Scanned PDFs: While OCR is supported, extremely low-quality scans (images of text without a text layer) may result in lower confidence scores.
• Language: Currently optimized for English and Italian reports. Support for German, French, and Spanish is in beta.

---

Support & Feedback

If you encounter any issues, have feature requests, or need a custom integration for your enterprise pipeline, please create an issue in the tab above.