Website Contact & Social Extractor

Business website URLs in → verified contact channels out. Feed it your lead list; get back emails, phones, social profiles, the contact page, and whether a contact form exists — clean, deduplicated, machine-readable.

What you get per website

• Emails — mailto links and page text, with aggressive false-positive filtering (no image@2x.png artefacts, no build hashes, no placeholder domains) and optional DNS MX validation so addresses whose domain cannot receive mail are dropped, not delivered to you as "leads"
• Phones — publisher-declared tel: links first, conservative pattern match as fallback, normalised
• Social profiles — Facebook, Instagram, LinkedIn, X/Twitter, YouTube, TikTok (share/plugin URLs excluded)
• Contact page URL and contact-form detection — so a site that publishes no email still gives your workflow a next step
• pagesChecked so you can verify exactly what was read

Design choices that matter

• Fail-open MX validation: only an authoritative "domain cannot receive mail" drops an email. A DNS hiccup flags the address emailsMxUnverified instead of silently destroying your data.
• Deterministic: plain HTTP fetch of home + contact/about pages. No headless browser, no LLM, no login, no anti-bot circumvention — it reads ordinary public pages the way any visitor does. Fast and cheap.
• Honest output: no email found means no email exists on those pages — not a guessed firstname@domain.com. Guessed emails bounce and burn your sender reputation.

Input

{ "websites": ["https://business-1.com", "https://business-2.com"], "maxPagesPerSite": 3, "validateMx": true }

Up to 500 websites per run.

Output example (per website)

{
  "website": "https://example-business.com/",
  "ok": true,
  "emails": [
    { "value": "office@example-business.com",
      "sourceUrl": "https://example-business.com/contact/",
      "dnsStatus": "domain_mail_capable" }
  ],
  "phones": [ { "value": "+15551234567", "sourceUrl": "https://example-business.com/" } ],
  "socials": { "facebook": { "urls": ["https://facebook.com/examplebiz"],
               "sourceUrl": "https://example-business.com/" } },
  "contactPageUrl": "https://example-business.com/contact/",
  "hasContactForm": true,
  "pagesChecked": ["https://example-business.com/", "https://example-business.com/contact/"]
}

Every contact carries the source page URL it was found on — you can always answer "where did this email come from?"

About dnsStatus: domain_mail_capable means the domain has MX records and can receive mail. It does not prove the specific mailbox exists — no tool can prove that without sending mail. domain_no_mx addresses are dropped. dns_unavailable means the check could not run; the address is kept and flagged rather than silently discarded.

What this Actor does NOT do

• It never guesses or generates addresses (no firstname@domain.com patterns) — guessed emails bounce and burn your sender reputation
• It does not access login-only data, bypass CAPTCHAs, or enrich from private databases
• It cannot extract contacts a site only renders via JavaScript, or from sites that block ordinary fetches — such sites return their status honestly rather than fabricated data
• Results depend entirely on what the supplied site publicly serves. Not every site returns contacts.

Pricing

$4 per 1,000 websites processed (pay-per-event, $0.004 per site). Process a 500-site lead list for $2. A site is charged when it has been fetched and analysed — including sites that legitimately return no contacts (the "no published email, has contact form" answer is itself the data).

FAQ

Why did a site return no emails? Because none are published on its checked pages. See hasContactForm and contactPageUrl for the next-best route, and pagesChecked for exactly what was read. Can it do thousands of sites? Up to 500 per run; queue multiple runs or schedule for larger lists. Does it use an LLM? No. Deterministic fetch + parse. That is why it is fast and cheap.

Use with

n8n, Make, Zapier, AI agents, or straight from the API. Clear input/output schema — agent-friendly.