Website Contact Extractor: Emails and Phone Numbers avatar

Website Contact Extractor: Emails and Phone Numbers

Pricing

from $2.10 / 1,000 result scrapeds

Go to Apify Store
Website Contact Extractor: Emails and Phone Numbers

Website Contact Extractor: Emails and Phone Numbers

Extract public email addresses and phone numbers from websites. Get structured results with explicit statuses for contacts found, empty results, failures and duplicates. The result event is charged only for sites with valid contacts.

Pricing

from $2.10 / 1,000 result scrapeds

Rating

0.0

(0)

Developer

Torix

Torix

Maintained by Community

Actor stats

0

Bookmarked

1

Total users

1

Monthly active users

3 days ago

Last modified

Features

Limited permissions

Share

Website Contact Details Extractor — Emails, Phones & Social Profiles from Any Website

Paste a list of websites or domains and get back one clean lead row per company: emails, phone numbers, LinkedIn, Facebook, Instagram, X/Twitter, YouTube, TikTok, WhatsApp and more, plus company name, address and the contact page URL.

The actor visits the homepage, finds the pages where contacts actually live (Contact, About, Imprint/Impressum, Team, Support, Locations, in 10+ languages) and merges everything into a single deduplicated record. It runs on plain HTTP with no headless browser, so it is fast and cheap: hundreds of websites per minute.

How to run it in 3 steps

  1. Add your input. In Websites or domains paste domains or homepage URLs, one per line — e.g. apify.com. The form is already filled with a working example, so you can skip this step for a first try.
  2. Press Start. The prefilled run returns 3 results in about 5 s and the actor's own charge is at most about $0.009 (only contacts_found rows are charged, see Row status and billing). Set Max results (cost cap) to cap the cost of bigger runs; you are charged only for websites with status: "contacts_found".
  3. Get the data. Open the Output tab and export JSON, CSV or Excel — or run it from the API, Make, Zapier, n8n or an AI agent (see below).

Features

  • One result row per website. Every website gets one result row with all of its emails, phones and socials, not one row per page. The dataset can also hold free note rows for the same domain (duplicate or a charge correction): see Row status and billing.
  • Pay per website with contacts, not per page. Page-based contact scrapers usually crawl 5–10 pages per site, which works out to about $0.01–0.02 per website. Here the actor's charge is $0.003 per website with a valid contact (status: "contacts_found"), however many pages were read; websites that fail or have no contacts do not trigger that charge.
  • Finds hidden emails: decodes Cloudflare email protection ([email protected]) and obfuscated addresses like info [at] acme [dot] com, and reads schema.org / JSON-LD business data.
  • Clean data: filters junk such as logo@2x.png, noreply@sentry.io, placeholder emails, form-placeholder phone numbers, dates and registration numbers. Phones are normalised and deduplicated (+1 800… and 1-800… count as one). Share buttons are not reported as social profiles.
  • Ranks the results: primaryEmail prefers an on-domain role inbox (info@, hello@, sales@…), and emailSources tells you which page each email came from.

Use cases

  • Enrich Google Maps / Yelp / directory exports that have a website column but no email.
  • Build cold-outreach lists from a list of company domains (CRM, LinkedIn Sales Navigator, Crunchbase, trade-show exhibitor lists).
  • Find social profiles of competitors, suppliers or influencers' businesses.
  • Clean and verify contact data in your CRM on a schedule.

Input

{
"websites": ["apify.com", "https://www.joesstonecrab.com", "hubspot.com"],
"maxPagesPerSite": 6,
"sameDomainEmailsOnly": false,
"includeFailedSites": true,
"maxConcurrency": 40
}
OptionDefaultNotes
websites—Domains or URLs, one per line. Duplicate domains are processed once.
datasetId—Optional. Dataset of another run (e.g. Google Maps leads): its websites are processed, no copy-paste. See Enrich a leads list below.
websiteFieldautoDataset column that holds the website (website, contact.site …). Empty = detect website, websiteUrl, domain, url …
maxPagesPerSite6Homepage + up to 5 best contact-related pages. 1 = homepage only.
sameDomainEmailsOnlyfalseKeep only emails on the website's own domain.
includeFailedSitestrueAdd status: "failed" rows with the reason (offline, 403, timeout…); they do not trigger the result charge.
maxResults0Cost cap: stop taking new websites after N charged contacts_found sites (0 = no limit). Websites already in progress may still finish.
maxConcurrency40Websites in parallel.
requestTimeoutSecs20Per page.
proxyConfigurationoffEnable Apify Proxy if many of your sites block cloud servers.

Enrich a Google Maps / any leads list

Already have a list of companies from another Actor? Skip the copy-paste:

  1. Run your lead source (for example a Google Maps scraper) and open the run's Storage tab → copy the Dataset ID.
  2. Here, paste it into Dataset of another run — and, if your column is not called website, enter its name in Website column.
  3. Set Max results (cost cap), press Start. Every distinct website in the dataset becomes one row with emails, phones and socials.
{ "datasetId": "<DATASET_ID>", "websiteField": "website", "maxResults": 500 }

Rows without a website are skipped, duplicate domains are processed once, and url is only used when it is not a Google Maps, Facebook, LinkedIn or Yelp page. If the dataset is empty, not yours or has no website column, the run stops at once with a message that says what to change — no charge. The three sample sites pre-filled in the form are ignored when a dataset is given. The dataset must belong to your account (or be public).

Output

Example output (synthetic)

All records below are synthetic illustrations of the row layout of this version, not output of a run: the websites, company names and emails are made up and values are shortened. Real values depend on each website.

A website with a valid contact (status: "contacts_found", the only billable status):

{
"website": "acme-corp.com",
"domain": "acme-corp.com",
"url": "https://acme-corp.com/",
"status": "contacts_found",
"billable": true,
"chargeEvent": "result",
"validation": "format-only",
"coverage": "full",
"pagesFailed": 0,
"companyName": "Acme Corp (made up)",
"title": "Acme Corp",
"description": "Synthetic description…",
"language": "en",
"address": null,
"primaryEmail": "info@acme-corp.com",
"primaryPhone": null,
"emails": ["info@acme-corp.com"],
"phones": [],
"emailSources": { "info@acme-corp.com": "https://acme-corp.com/contact" },
"linkedin": null,
"facebook": null,
"instagram": null,
"twitter": null,
"youtube": null,
"tiktok": null,
"socials": {},
"contactPageUrl": "https://acme-corp.com/contact",
"pagesScraped": 2,
"scrapedAt": "2026-01-01T00:00:00.000Z"
}

A website that was read but has no valid contact (status: "no_contacts", not billable; coverage: "partial" here means one extra page could not be read):

{ "website": "quiet-shop-demo.com", "domain": "quiet-shop-demo.com", "status": "no_contacts", "billable": false, "chargeEvent": null, "validation": "format-only", "coverage": "partial", "pagesFailed": 1, "primaryEmail": null, "primaryPhone": null, "emails": [], "phones": [] }

If a billable row was saved but its charge is not confirmed, a correction row is added; the earlier contacts_found row stays in the dataset:

{ "website": "acme-corp.com", "domain": "acme-corp.com", "status": "charge_unconfirmed", "billable": false, "chargeEvent": null, "validation": "format-only", "emails": [], "phones": [], "scrapedAt": "2026-01-01T00:00:00.000Z", "error": "run restarted after the row was saved; charge not retried", "corrects": "billable row of this domain" }

socials can also contain pinterest, github, telegram, whatsapp, threads, discord and yelp. The flat linkedin/facebook/… columns make CSV and Excel exports easy to use.

A website that could not be scraped (synthetic; failed rows do not trigger the result charge):

{ "website": "old-shop.com", "domain": "old-shop.com", "status": "failed", "billable": false, "chargeEvent": null, "error": "HTTP 403 (blocked — try enabling a proxy)", "httpStatus": 403, "emails": [], "phones": [], "scrapedAt": "2026-01-01T00:00:00.000Z" }

Ask in plain English

Using Apify AI, Claude, ChatGPT or another MCP client? Just describe what you need, the agent fills the input for you:

You askInput the agent sends
"Find the contact email and phone of apify.com and stripe.com"{"websites": ["apify.com", "stripe.com"]}
"Get LinkedIn and Instagram links for allbirds.com, gymshark.com and glossier.com"{"websites": ["allbirds.com", "gymshark.com", "glossier.com"]}
"Add emails to the websites in my Google Maps dataset abc123"{"datasetId": "abc123", "maxResults": 500}

Use with AI agents (MCP), LangChain, RAG & the API

This actor is a ready-made tool for AI agents: an LLM can call it, wait for the run and read clean JSON back.

MCP server (Claude Desktop, Claude Code, Cursor, VS Code, any MCP client) — add Apify's hosted MCP server with this actor enabled as a tool:

{
"mcpServers": {
"apify": {
"url": "https://mcp.apify.com/?tools=torix33/website-contact-details-extractor",
"headers": { "Authorization": "Bearer <APIFY_TOKEN>" }
}
}
}

LangChain / LangGraph (pip install langchain-apify, set APIFY_API_TOKEN):

import json
from langchain_apify import ApifyActorsTool, ApifyWrapper
from langchain_core.documents import Document
# 1) give the actor to an agent as a tool
tool = ApifyActorsTool("torix33/website-contact-details-extractor")
# 2) or load results as Documents for a RAG index
loader = ApifyWrapper().call_actor(
actor_id="torix33/website-contact-details-extractor",
run_input={"websites": ["apify.com", "https://www.joesstonecrab.com", "hubspot.com"], "maxPagesPerSite": 6, "sameDomainEmailsOnly": False, "includeFailedSites": True, "maxConcurrency": 40},
dataset_mapping_function=lambda item: Document(page_content=json.dumps(item, ensure_ascii=False), metadata={"url": item.get("url")}),
)
docs = loader.load()

REST API — one synchronous call, dataset items in the response:

curl -X POST "https://api.apify.com/v2/acts/torix33~website-contact-details-extractor/run-sync-get-dataset-items?token=<APIFY_TOKEN>" \
-H "Content-Type: application/json" \
-d '{"websites": ["apify.com", "https://www.joesstonecrab.com", "hubspot.com"], "maxPagesPerSite": 6, "sameDomainEmailsOnly": false, "includeFailedSites": true, "maxConcurrency": 40}'

Also works with the Apify Python/JS clients, Make, Zapier, n8n, Google Sheets, webhooks and schedules. Export as JSON, CSV, Excel, XML or RSS.

RAG tip: An agent can enrich a lead list on the fly: send the domains, read primaryEmail, phones and socials from the rows and write them back to your CRM.

Lead-generation workflow: works well with

The same people who use this actor usually need one of these next:

Pricing

  • The prefilled input covers 3 websites → at most $0.009 of result charges, and only for websites that return contacts_found.

$3.00 per 1,000 websites with a valid contact ($0.003 per website with status: "contacts_found"). All pages crawled on a site are included. The result event is not charged for failed, no_contacts or duplicate rows, or for correction rows. Only events listed on this Actor's pricing tab can be charged.

ExampleHighest result cost, if every website has contacts
500 restaurant websites from a Google Maps exportup to $1.50
10,000 company domains from a CRMup to $30.00

Set Max results to cap the number of billable websites per run. When Max results or your spending limit is reached, the actor stops taking new websites. Websites already in progress may still be saved; whether their charge is applied is decided by the platform's limit, and if it is not, a charge_not_applied note row is written.

Row status and billing

Every row has a status, a billable flag and a chargeEvent ("result" on billable rows, otherwise null). Contacts are checked for format only (validation: "format-only"): nothing is sent to an address or number to verify it.

statusMeaningbillable
contacts_foundThe website was read and has at least one valid contact: a phone number with a valid format, or an email on the website's own domain.true
no_contactsThe website was read but has no such contact. Emails of other domains may still be listed in emails; they do not make a row billable.false
failedThe website could not be read (offline, blocked, not HTML, or still failing after the retries below). See error and httpStatus.false
duplicateAnother input of the same run already ended on this domain (duplicateOf).false
charge_not_appliedCorrection row: the contacts_found row of this domain was saved but nothing was charged (for example the result event is not in the run's pricing, or the platform reported 0 charged events).false
charge_skippedCorrection row: saved, but the actor could not record its billing state or read the pricing, so it did not charge.false
charge_unconfirmedCorrection row: saved, but the charge call failed or the run restarted before the result was known. The charge is not retried and may or may not have been applied.false
  • A correction row has corrects set and the reason in error. The earlier contacts_found row of that domain stays in the dataset (datasets are append-only). A contacts_found row with no correction row is normally one that was charged, but the Apify billing details of the run, not the dataset, are the authority on what was charged.
  • Saving a row and charging for it are two separate calls, not one atomic operation. The actor saves the row first and charges afterwards, at most once per website per run. It does not guarantee that a saved row is charged or that a charge succeeds. After a restart it does not charge a website it has already charged or saved, as far as its per-website record in the run's key-value store (BILLING-<domain>) can be read.
  • coverage: "partial" means at least one extra page of the website failed (error, non-200 or non-HTML response). A partial contacts_found row is still billable; a partial no_contacts row may have missed contacts.
  • Retries: network errors, HTTP 429 and HTTP 5xx are retried up to 2 more times (3 attempts); an address that does not resolve, refuses the connection or fails its certificate check is tried at most twice; other HTTP errors (403, 404 …) and non-HTML pages are not retried; a website that still fails becomes a failed row. A failed dataset save is retried together with the website. The charge itself is never retried automatically, because its outcome may be unknown.
  • A charge is attempted only if the result event is registered in this Actor's pay-per-event pricing on the platform. If it is not, rows are saved without any charge and a charge_not_applied note row is written.
  • The run's status message and the RUN_STATS record count charged, no-contacts, failed and duplicate websites and charge problems. They are the actor's own counters, not a billing statement.

Speed

Benchmark of an earlier version on the Apify platform (before the billing bookkeeping described above was added; not re-measured): 300 real company websites (≈1,330 pages) in 77 seconds with the default 2 GB of memory, or about 230 websites per minute. Every website is a different server, so you can safely raise maxConcurrency (and memory) for big lists.

Tips

  • Phone numbers are returned in the format the site uses, cleaned to digits with an optional leading +. Numbers without a country code are local to the website's country.
  • Sites that need JavaScript for their menus are still covered: if no contact links are found in the HTML, the actor tries the usual /contact, /contact-us, /about and /impressum paths.
  • Blocked sites (403): a few websites block all cloud servers. Enable Apify Proxy and re-run only the failed rows. Failed rows do not trigger the result charge, but a re-run is a new run with its own charges for every website that returns contacts.

Limitations

  • Only contacts a website publishes (contact, about, imprint pages); no guessed or verified emails.
  • Sites that render content only with JavaScript or block bots (Cloudflare challenge) may fail or return no_contacts; these rows do not trigger the result charge.
  • Phone numbers are normalised from page text; numbers in images are not read.
  • Contacts are format-checked only: an email or phone that looks valid can still be wrong, outdated or unmonitored.
  • Billing is not atomic and not guaranteed per row: a billable row can end as charge_not_applied, charge_skipped or charge_unconfirmed (see Row status and billing).

FAQ

Do I pay per page or per website? Per website with a valid contact (status: "contacts_found"). All pages crawled on a site are included in one $0.003 result event. Websites that fail, have no contacts or are duplicates do not trigger that event.

Am I charged if a website has no contacts, or if a charge is unclear? Not by the result event: no_contacts, failed and duplicate rows are not billable. If a billable row was saved but its charge did not go through or is unknown, a correction row (charge_not_applied, charge_skipped or charge_unconfirmed) says so; see Row status and billing. The Apify billing details of the run are the authority on what was charged.

Can it find emails hidden behind Cloudflare email protection? Yes. [email protected] links and obfuscated addresses like info [at] acme [dot] com are decoded automatically.

Where do I get the list of websites? Any CSV export works: a CRM, a Google Maps scraper, a directory. Paste the domains or URLs into websites.

Is it an alternative to Hunter.io or Snov.io? For company contact pages, yes: it reads what the website itself publishes (contact, about, imprint pages) instead of guessing addresses.

The actor only collects information that the websites publish themselves. Personal data (e.g. named employee emails) is protected by laws such as GDPR and CAN-SPAM in many jurisdictions. Make sure you have a legitimate reason to process it and that your outreach complies with those laws.

Support

Missing a field, or a site that should work but doesn't? Open an issue on the Issues tab with the domain, and fixes usually ship within days.