An email scraper API takes a domain and returns everything contactable that the site publishes about itself — emails, phone numbers, social profiles and postal addresses — merged into one record per domain.
The word merged is the important one. It does not read the homepage and stop. For each domain it loads the homepage, follows the site's own links to contact, about and legal pages, falls back to the sitemap when those links are missing, and combines everything it finds across all of them: mailto: and tel: links, JSON-LD Organization and LocalBusiness blocks, and page text. One domain in, one complete record out.
$0.02 per delivered site, up to 200 sites per run. Nothing delivered means nothing charged.
Post an array of domains, up to 20 per run. Each one gets a bounded crawl: homepage first, then the contact-shaped pages discovered from the site's own navigation, with the sitemap as a fallback when a site has no obvious contact link. max_pages caps how far it goes.
Three extraction paths run in parallel on every page, and using all three is what separates a useful result from an empty one:
mailto: and tel: hrefs are the highest-confidence source — they are machine-readable by design.Organization and LocalBusiness blocks frequently carry contact points and postal addresses that never appear as visible text.The German /impressum case is worth calling out: legally mandated disclosure pages in several European countries carry complete contact details, and a crawler that only looks for /contact misses them entirely.
domains is required: an array of domains or URLs. max_pages bounds the crawl per domain — the lever between coverage and cost, though the price is per site rather than per page, so a higher budget improves the result without changing the bill. country sets the exit region, which matters for sites that geo-vary their content. max_results caps how many domains the run processes.
One record per domain, not one per contact. emails, phones, socials and addresses are arrays holding everything found across every page scanned. contact_pages lists the URLs that produced them and pages_scanned how many were read — together they make any result auditable rather than something you take on faith. status distinguishes a site with no published contacts from one that failed to load, which are different facts. site_name is the organisation name where the site declares one.
$0.02 per delivered site ($20 per 1,000). Nothing delivered means nothing charged, and the $2 monthly free credit covers roughly 100 sites here. Volume tiers take up to 30% off.
$0.02 per site, $20 per 1,000 — the second-highest unit in the catalogue, and it is priced per site rather than per page or per contact. A domain that takes four page fetches costs the same as one that takes two, so raising max_pages improves coverage without increasing what you pay.
Enriching 1,000 domains is $20. The free $2 monthly allowance covers about 100 sites.
The email-finder tools solve a different problem, and confusing the two wastes money. They answer "what is this named person's email at this company", usually by pattern inference plus verification against a database of previously seen addresses. That is genuinely hard and worth paying for when you need to reach a specific individual.
This answers "what contact details does this organisation publish". It finds hello@, support@, sales@, the switchboard number, the registered address — the front door rather than a named person. For local businesses, SMBs and any company without a public org chart, the front door is frequently the only door there is, and it is also the address the company chose to publish, which is a materially different consent posture from an inferred personal address.
They are complementary. Use this to find the organisation's published contacts; use a finder when you need to reach a specific person by name.
A regex over a homepage takes ten minutes and gets you maybe a third of the contacts that are actually there. What is missing:
name [at] domain [dot] com. Naive regex misses them and sometimes invents them.Attach published contact details to a list of domains you already hold, with contact_pages as the audit trail for each one.
Contact routes for organisations you need to reach, gathered from what they publish themselves.
Registered addresses and legal-page contacts, particularly in jurisdictions with mandatory disclosure pages.
Pair with the company profile collector for firmographics alongside the contact layer.
Up to 200 domains per run, with the per-domain crawl bounded by max_pages. Contacts published only behind a form, or rendered as an image, will not be found — pages_scanned and contact_pages tell you what was actually read, so a thin result is explainable rather than mysterious.
Legally, separate the two questions. Collecting contact details a company publishes on its own website is generally lawful in most jurisdictions and is about as consent-adjacent as public data gets — the organisation chose to publish it. Using those addresses for unsolicited marketing is governed independently by GDPR and PECR in Europe, CAN-SPAM in the US and CASL in Canada, and some of those require consent before the first message. A named individual's work address (firstname@) is personal data even on a corporate site, in a way that info@ largely is not. Get advice on the sending, not just the collecting. Not legal advice.
There is a free allowance of $2 a month with no card, about 100 sites at $0.02 each. Free extractors on GitHub are usually a regex over the homepage, which finds a fraction of what a site actually publishes — the contacts are mostly on pages the homepage links to.
$0.02 per site — $20 per 1,000 — priced per domain rather than per page or per contact. A site needing four page fetches costs the same as one needing two, so raising max_pages improves coverage without increasing the bill.
Different questions. Finders answer 'what is this named person's address at this company', usually by pattern inference and verification. This answers 'what contact details does this organisation publish' — the front door rather than a named person. They complement each other rather than compete.
It reads mailto: and tel: links, JSON-LD Organization and LocalBusiness blocks, and page text. Addresses assembled by client-side JavaScript at runtime or rendered as images are not recovered — and pages_scanned plus contact_pages tell you exactly what was read, so a thin result is explainable rather than a mystery.
As many as max_pages allows: the homepage first, then contact, about and legal pages found from the site's own links, with the sitemap as a fallback. Following the site's navigation beats guessing paths — contact pages live at /impressum, /kontakt and /contatti as often as at /contact.
That is a separate question from whether collection was lawful, and the answer depends on where you and the recipient are. GDPR and PECR in Europe, CAN-SPAM in the US and CASL in Canada all govern unsolicited commercial messages, and some require consent before the first send. A named individual's address is personal data even on a corporate site. Take advice before running a campaign.
You get a record with empty arrays and a status explaining what happened, plus the list of pages that were scanned. That distinguishes 'this site publishes no contacts' from 'this site failed to load', which are different findings and should not look the same.
Skip the marketing. QuantumProxies.io publishes a machine-readable site map, Markdown for every page, a free MCP server, and APIs billed from the same balance as your proxies.
Connect in one command: npx -y quantumproxies-mcp