Business Website Contact & Lead Extractor

Turn public company websites into structured, reviewable business-contact records. The Actor scans each supplied homepage and prioritizes same-site pages such as Contact, About, Team, Support, Imprint, and Locations.

What it extracts

• Public email addresses with role/named classification, domain-match evidence, source pages, and free-provider labels
• Phone numbers from tel: links and carefully filtered visible text
• LinkedIn, Facebook, Instagram, X/Twitter, YouTube, TikTok, and GitHub profiles
• Postal addresses explicitly published in Schema.org JSON-LD
• Contact forms, company name signals, processed pages, and page-level errors
• A conservative contactConfidence label based only on collected evidence

The Actor does not guess email addresses, enrich personal identities, bypass logins, or scrape private pages. Off-domain/free-provider named-email collection is off by default; public named emails on the company's own domain can still be returned.

Quick start

{
  "startUrls": [
    {"url": "https://www.iana.org/contact"},
    {"url": "https://example.com/"}
  ],
  "maxWebsites": 2,
  "maxPagesPerWebsite": 5,
  "maxDepth": 1,
  "includePersonalEmails": false,
  "timeoutSeconds": 20
}

Click Start to run the included example, then open Output. One dataset record is written per successfully scanned website. The table includes copyable emailAddresses and phoneNumbers; the detailed emails and phones arrays include evidence and source URLs. Export the dataset as JSON, CSV, or Excel.

OUTPUT in the default key-value store contains the run summary and errors for websites that produced no usable page. A run in which every website fails is marked Failed, with the reasons preserved in OUTPUT. A website may scan successfully but contain no contact details.

Safety and predictable cost

• Public HTTP(S) URLs and standard ports only
• DNS/IP checks block localhost, private, link-local, reserved, and other non-global targets; connections use a validated address while preserving HTTPS certificate verification
• Redirect destinations are revalidated; maximum five redirects
• robots.txt is respected; unverifiable robots rules fail closed
• At least one second between requests to the same origin
• Maximum 2 MB per HTML response and 512 KB for robots.txt
• Maximum 50 websites, 10 page attempts per website (failed requests count), link depth 2, and roughly 4.5 minutes per run
• Pay-per-event budget is checked before scanning, and the SDK saves and charges each record together; the last affordable record is retained

Start with one or two websites and a spending limit. Increase limits only after reviewing data quality.

Output notes

contactConfidence means evidence density, not commercial quality or deliverability:

• high: at least five pieces of public contact evidence
• medium: two to four
• low: one
• none: no contact evidence on successfully scanned pages

Email and phone counts are uncapped totals; saved detail arrays may be truncated according to input limits. Every email and phone includes the page URLs where it was observed.

In large batches, the run time budget may stop processing before every website is scanned. Check unprocessedWebsites and stoppedByTimeBudget in OUTPUT, then run the remaining URLs in a new batch. Invalid URLs produce an explicit input error instead of being silently dropped.

Pricing

When monetization is enabled, the configured price is $0.005 per successfully scanned website ($5 per 1,000 websites). A website scan includes up to your configured page limit. Failed websites produce no charged result. Successfully scanned sites without publicly available contacts still count as a website scan. Always check the live Pricing tab before running.

Limitations

• Server-rendered HTML only; JavaScript-only content is not rendered
• Contact data can be outdated or intentionally absent
• Phone formats vary globally, so ambiguous digit strings are deliberately omitted
• Addresses are returned only from explicit structured data to avoid inventing locations
• Domain matching is a transparent hostname comparison, not a legal or ownership assertion
• Users remain responsible for website terms, privacy law, marketing consent, and permitted use

For a reproducible support report, include the public URL, run ID, expected field, and actual field. Do not include passwords, private pages, or sensitive personal data.