Extract Website Contacts for B2B Lead Generation (2026)

Thirdwatch's Website Contact Scraper turns a list of company domains into structured contact data — emails, phone numbers, social profile links, and a postal address — read from each company's own public website. It crawls the homepage plus contact, about, team, and imprint pages, deobfuscates (at)/(dot) emails, and returns one clean row per domain. Built for B2B lead generation, sales prospecting, and CRM enrichment teams who start from a domain list and need contact points without manual page-clicking.

Why scrape company websites for lead generation

A company's own website is the most authoritative public source of its contact details — and the one your prospects keep current. Before any outreach, a sales rep ends up on the same pages anyway: homepage footer, /contact, /about, /team, the imprint or legal page. The job-to-be-done is mechanical: visit those pages, copy the support and press emails, grab the phone number, note the LinkedIn and Twitter links, and read off the head-office address. Done by hand it scales to maybe 20-30 domains an analyst-hour, and the copy-paste step is where transcription errors creep into your CRM.

That manual cost is why so much pipeline data is stale. According to HubSpot's guidance on CRM data hygiene, B2B contact databases degrade by roughly 22-30% a year as people change roles and companies rebrand, so records need continual refreshing. Re-deriving contact points straight from the live website is the cheapest way to keep a list fresh. Feed the actor a column of domains — from a trade-show export, a Google Maps sweep, or an existing CRM segment — and it returns the same five things a rep would have copied by hand: emails, phones, socials, address, and a pagesCrawled count, as clean JSON ready for pandas, HubSpot, or Salesforce.

How does this compare to the alternatives?

Three common ways to get contact data off company websites at scale:

A DIY crawler looks easy until you hit obfuscated emails, tel: parsing, asset false-positives like logo@2x.png, and the first site that 403s your script. Email-only finder APIs guess firstname.lastname@ patterns and ignore phones, socials, and addresses entirely. The Website Contact Scraper actor page returns all four contact types from what the site actually publishes, with the obfuscation handling and anti-bot fallback already solved.

How to extract website contacts for lead generation in 4 steps

Step 1: How do I authenticate against Apify?

Sign in at apify.com (free tier, no credit card), open Settings → Integrations, and copy your personal API token. Every example below assumes the token is in APIFY_TOKEN:

export APIFY_TOKEN="apify_api_xxxxxxxxxxxxxxxx"

Step 2: How do I enrich a single domain?

Pass one or more domains in the domains array. Bare domains, full URLs, and www. prefixes all work — each entry produces one result row.

import os, requests, pandas as pd

ACTOR = "thirdwatch~website-contact-scraper"
TOKEN = os.environ["APIFY_TOKEN"]

resp = requests.post(
    f"https://api.apify.com/v2/acts/{ACTOR}/run-sync-get-dataset-items",
    params={"token": TOKEN},
    json={
        "domains": ["basecamp.com"],
        "maxPagesPerDomain": 10,
    },
    timeout=300,
)
row = resp.json()[0]
print(row["emails"], row["phones"], row["socials"], row["address"])

A single-domain enrichment finishes in well under a minute. maxPagesPerDomain controls depth: leave it at the default 10, drop it to 3 for a fast homepage-plus-contact-page pass, or raise it toward 30 for thorough coverage of sprawling sites.

Step 3: How do I build a prospecting list from a domain list?

Pass the whole list in one run. The actor crawls each domain independently and returns one row per domain in the dataset.

domains = (
    pd.read_csv("tradeshow_exhibitors.csv")["website"]
    .dropna()
    .str.strip()
    .tolist()
)

resp = requests.post(
    f"https://api.apify.com/v2/acts/{ACTOR}/run-sync-get-dataset-items",
    params={"token": TOKEN},
    json={"domains": domains, "maxPagesPerDomain": 8},
    timeout=3600,
)
df = pd.DataFrame(resp.json())
print(f"{len(df)} domains, "
      f"{df.emails.map(bool).sum()} with email, "
      f"{df.phones.map(bool).sum()} with phone")

For very large lists, split into runs of a few hundred domains and append the datasets — that keeps each run inside the sync timeout and makes failures easy to retry per batch.

Step 4: How do I push enriched contacts to HubSpot?

Filter to rows that have an actionable contact point, then map each to your CRM's Company schema and POST. The socials field is an object keyed by platform, so pull individual networks by key.

import requests as r
HUBSPOT_TOKEN = os.environ["HUBSPOT_TOKEN"]

QUALIFIED = df[df.emails.map(bool) | df.phones.map(bool)].copy()

for _, row in QUALIFIED.iterrows():
    socials = row["socials"] or {}
    r.post(
        "https://api.hubspot.com/crm/v3/objects/companies",
        headers={"Authorization": f"Bearer {HUBSPOT_TOKEN}"},
        json={"properties": {
            "domain": row["domain"],
            "email": (row["emails"] or [""])[0],
            "phone": (row["phones"] or [""])[0],
            "address": row["address"] or "",
            "linkedin_company_page": socials.get("linkedin", ""),
            "twitterhandle": socials.get("twitter", ""),
            "lifecyclestage": "lead",
            "source": "Website Contact Scraper",
        }},
        timeout=10,
    )
print(f"{len(QUALIFIED)} enriched companies pushed to HubSpot")

Schedule the run on Apify's scheduler and forward each completed dataset to your CRM with a webhook for a self-maintaining enrichment loop.

Sample output

Each dataset row is one domain's contact profile. Fields mirror the actor's dataset exactly — domain, emails, phones, socials, address, pagesCrawled, and sourceUrls. Two redacted records:

[
  {
    "domain": "basecamp.com",
    "emails": ["support@basecamp.com", "press@basecamp.com"],
    "phones": ["+1-312-261-9899"],
    "socials": {
      "twitter": "https://twitter.com/basecamp",
      "youtube": "https://youtube.com/basecamp",
      "github": "https://github.com/basecamp"
    },
    "address": "30 N Racine Ave, Chicago, IL, 60607, US",
    "pagesCrawled": 6,
    "sourceUrls": {
      "address": "https://basecamp.com/about"
    }
  },
  {
    "domain": "examplestudio.io",
    "emails": ["hello@examplestudio.io"],
    "phones": [],
    "socials": {
      "linkedin": "https://www.linkedin.com/company/examplestudio"
    },
    "address": null,
    "pagesCrawled": 4
  }
]

domain is the normalized key (www stripped) — use it for dedup and CRM upsert. emails and phones are deduplicated arrays, so a company with no published phone returns [] rather than a guess. socials is an object keyed by platform (linkedin, twitter, facebook, instagram, youtube, tiktok, github); read individual networks by key. address is best when the site exposes structured data and may be null otherwise. sourceUrls tells you exactly which page each derived field came from — handy for auditing the second record above, where no address was published at all.

Common pitfalls

Three things to plan for in website-contact pipelines. Coverage is never 100% — some sites publish no email at all and route everything through a contact form, so the emails array can come back empty; treat phones and socials as fallback contact points and don't filter the whole row out on missing email. JavaScript-only contact widgets — emails rendered entirely client-side may not appear in a pure-HTTP crawl, which is the right trade-off for speed and cost on the vast majority of sites that ship contact details in HTML. Format drift — phones arrive in each site's own display format and addresses in their natural form, so normalize phones to a single digit string before handing them to a dialer and parse addresses downstream rather than expecting clean city/state columns.

Thirdwatch handles each of these at the crawl layer: obfuscated (at)/(dot) emails are deobfuscated, asset false-positives like logo@2x.png are filtered out, social share and tracking widgets are dropped so only real profile links survive, and a 403 or anti-bot challenge triggers an automatic retry with built-in anti-bot handling. You also stay on the right side of the rules — the actor only reads contact details a company publishes on its own public website, never private or gated data, so always respect each site's terms and applicable anti-spam and data-protection law, such as the FTC's CAN-SPAM rules for outreach, when using what you extract.

Related use cases

• Scrape Google Maps businesses for lead generation — get the domain list this actor enriches
• Scrape JustDial for local business leads in India
• Enrich your CRM with LinkedIn profile data
• Build account-based marketing from LinkedIn companies
• The complete guide to scraping business data
• All Thirdwatch use-case guides