Website email scraper

The email address, phone number, contact form, social accounts and postal address each website on a list publishes, as one row per site in a CSV. Give it a file of websites, one per line, the way they sit in a spreadsheet column; bare hosts like rjf.uk.com work.

These scripts call Website Contact Scraper, which reads the page you give plus up to four more of that site's pages, the ones likeliest to carry contact details (the contact, about or team page).

The write-up, with a real run on 36 accountancy firms in Manchester, is How to find the email address for a list of company websites.

Run it

You need an Apify account and its API token, from Console → Settings → API & Integrations.

Node (18 or newer):

cd node
npm install
APIFY_TOKEN=your_token COUNTRY=GB node contacts.mjs ../websites.txt

Python (3.11 or newer):

cd python
python -m venv .venv && . .venv/bin/activate
pip install -r requirements.txt
APIFY_TOKEN=your_token COUNTRY=GB python contacts.py ../websites.txt

websites.txt holds three of the firms from the post. Put your own list in it, or pass websites as arguments instead of the file. COUNTRY is the two-letter code used to read phone numbers written without a country code ("0161 …"); leave it out and the scraper works it out from each site. MAX_PAGES (default 5) is how many pages of each site may be read; MAX_PAGES=1 reads only the page you gave, which costs less and finds fewer addresses. Both scripts print one line per site and write contacts.csv:

  taylorevans.co.uk  info@taylorevans.co.uk  +441613021571
  ac-accounts.co.uk  info@ac-accounts.co.uk  +441619621855
  rjf.uk.com  hello@rjf.uk.com  +441615040629
3 of 3 site(s) read, 3 with an email -> contacts.csv

What a row looks like

sample-output.json holds four real rows from the run: a firm whose email was on its contact page, one whose email was on its home page, one that publishes no email at all, and a site that answered with a bot check. The fields you will use most:

Field	What it holds
email	The single best address: the site's own domain first, a general mailbox (info@, hello@) before a person's
phone	The single best number, in E.164 (+441615040629)
emails, phones	Everything found, each with the page it was found on
contactFormUrl	The page with the site's contact form, for firms that publish no address
social	One URL per platform: social.linkedin, social.facebook, social.instagram...
address	The postal address, when the site gives a real one; address.raw is the text
pagesRead, pagesReadCount	Which pages were read, and how many were charged

A site that could not be read has no emails key at all, and carries error and errorDescription instead. A site that was read and publishes no address has emails: []. Both scripts print the first kind and leave it out of the CSV; the run itself stays green either way.

Cost

$0.001 per run plus $0.00075 per page read, Apify's platform costs included. A page that turns out not to exist is not charged, and neither is a site that could not be read. On the 39 Manchester addresses, reading the home page only was 37 pages, $0.029; up to five pages was 161 pages, $0.122.

Limits

• Some sites answer with a bot check; those come back as challenge errors. Switching the scraper's proxy input to residential is the documented fix.
• Some firms publish no email anywhere: 4 of the 36 in the post. They usually have a phone number or a contact form instead.
• The emails list holds every address the pages printed, including other organisations' addresses and staff members' own. Use email for the one to write to, and check your local rules before contacting a named person.
• It reads up to maxPagesPerSite pages (at most 20), never a whole site.