Decision Maker Email Finder: Domain to Verified Contacts

Type a company domain. Get the people who decide there: name, title, the page that shows them at the company, a confidence grade, and a work e-mail the mail server accepted. It reads the company's own site, Wikidata, SEC filings and (with an operator key) the UK company register in one run, then checks the e-mail address, profiles the company and scores every person. An AI model reads each title and says whether that person decides on a purchase, and why.

No LinkedIn login, no cookies, no setup: paste domains and run. Unofficial: not affiliated with any company named in the results.

Free vs Paid

	Free plan	Paid plan
Domains per run	5	up to 1,000
Decision-makers per domain	3 (best score first)	up to 25 (default 25)
Verified e-mail rows	for the people shown	for the first 10 people per domain (the rest get guessed addresses only)
Company profile and buying-committee rows	up to 1 per domain	every one that qualifies
AI judgment rows	1 sample per domain	every person
Pick functions (roleFamilies) and "what changed since my last run" (previousDatasetId)	locked (paid plans only)	yes
Price	list price x 1.5	list price (the same for every paid plan, no volume tiers)

A free run ends with free-plan-locked rows (for example 3 of 8 decision-makers shown for hubspot.com) and one free-plan-preview row that says what the same input returns on a paid plan, with an estimate in dollars: the hidden people, their AI readings, the locked domains beyond the first 5 at this run's average, and the whole input on a paid plan. It also says the same rows cost 33% less on a paid plan. Estimates are labelled as estimates; the counts are measured. Free runs finish as SUCCEEDED; nothing looks like an error. On our 5-domain sample a paid run returned 2.4x the billable rows of the free run (74 vs 31).

What you pay (pay per event)

List price, what BRONZE and every higher plan pays. The free plan pays 1.5x.

Event	Charged for	Price
run-started	one run (also when every input line is unusable)	$0.005
domain-checked	one domain we checked: the sources answered, and the row names the company or says found: false (parked page, no website, not in a register)	$0.005
decision-maker-found	one person: name, title, source URL that shows them at the company, identity medium or high	$0.012
verified-email-found	one work e-mail the mail server accepted, on a domain that does not accept everything	$0.06
company-profile-enriched	one company profile with at least three of: tech, social, domain age, public contacts	$0.02
committee-mapped	one buying-committee map (three or more different functions found)	$0.02
change-detected	one person who is new, left or changed title since previousDatasetId	$0.01
decision-maker-judged	one AI judgment row (is this person a decision-maker, and why)	$0.004

Not charged: notice rows, a second source for the same person, a name without a title, low identity grade, guessed e-mail patterns, catch-all domains, role addresses (info@, sales@), and a domain whose sources did not answer (a timeout or failure on our side). A person unchanged since your earlier run is not charged again.

Example, one domain with three people and one verified e-mail: $0.005 + $0.005 + 3 x $0.012 + $0.06 + 3 x $0.004 = $0.118, plus a profile or committee row when the data is there. Over our 20-domain sample the average came to about $0.10 per domain with every layer on.

Expect this: verified e-mails are found on a minority of domains (catch-all domains cannot be confirmed). In the 20-domain sample 4 of 73 people had one; the other people carry up to five guessed addresses on their row, free. You pay $0.06 only when the mail server accepted the address.

Every input you typed is accounted for. An entry that was not looked up (wrong format, a repeat, over the plan's domain limit) is listed in one free input-skipped row (when none of your entries is usable, the same skipped list is on the single invalid-input notice instead); if the run stops early (time budget or your spending limit) the entries it did not reach are listed in one free not-processed row.

What you get

1. Collect: the company site (home, about, team, leadership, imprint, contact; links two levels deep; one sitemap hit), Wikidata (current CEO and chair, company IDs), SEC Form 3/4 (officers and directors of listed companies), RDAP (domain age). UK Companies House officers for .uk domains when the operator key is set.
2. Cross-check: the same person written differently in two sources becomes one row (Kate = Kathryn); for UK Companies House officers the name must appear on a company page (Wikidata and SEC people need no page match); customer quotes, former staff and outside board seats are filtered out. Identity grade high, medium or low.
3. Enrich: five work-address patterns checked against the mail server (MX and SMTP RCPT, no mail is sent), catch-all detection, role and disposable addresses; company profile with tech stack, socials and public contacts.
4. Judge: fit_score, send_decision (send, review or skip with reasons), confidence, and a one-line evidence_reason per person; a buying committee per company; changes against your earlier run; an AI judgment row per person with authority_level and buying_role.

Every row carries source_url, checked_at, confidence and basis, so a person or an AI agent can see where it came from and why.

Input

• domains (required): one per line. A web domain (acme.com), a website address, or a company name.
• maxPeoplePerDomain: 1-25, default 25 (free plan: 3).
• aiJudgment: default on. buyerContext: one sentence such as "the person who buys marketing software" to judge against.
• Paid plans: roleFamilies (exec, sales, marketing, ops, finance, tech, hr, legal) and previousDatasetId.
• proxyConfiguration: used for a site only after the direct request was refused.

Output

One row per company, person, e-mail, profile, committee, change or AI judgment (row_type). A person row:

{
  "row_type": "decision-maker",
  "domain": "buffer.com",
  "full_name": "Joel Gascoigne",
  "job_title": "CEO & Co-Founder",
  "role_family": "exec",
  "seniority": "c-level",
  "email": "joel@buffer.com",
  "email_status": "verified",
  "email_basis": "smtp-accepted",
  "catch_all": false,
  "identity_status": "medium",
  "fit_score": 72,
  "send_decision": {"action": "send", "risk_level": "medium", "confidence": 0.7},
  "evidence_reason": "Joel Gascoigne is listed as CEO & Co-Founder at Buffer by the company's own page buffer.com; one source only.",
  "source_url": "https://buffer.com/about",
  "confidence": 0.7,
  "basis": "Named with this title by: site."
}

The company row has sources_status (ok, none, failed or timeout per source) and partial: a count that is missing is unknown, not zero.

What is still being checked

• UK Companies House: built and tested with recorded answers; the live register answer has not been seen yet (checking). It runs only for .uk domains and only when the operator sets a key; without it the source is skipped and says so in sources_status.
• LinkedIn profile address by public search: tested with recorded answers; no live case where it returned an address has been seen (checking). When it finds an address the row says linkedin_source: public-search; when it does not, the field is empty and nothing is charged for it.

Notes

• Titles come from public pages and can be stale; always read source_url and identity_status before you send mail. On our 20-domain hand check 51 of 53 people were right, and the grade, source and reason are on every row so you can filter.
• Accuracy depends on what the company publishes. A domain with no public team page returns the company row and few or no people.
• You are responsible for how you use the contacts (e-mail, privacy and anti-spam rules where you send). The Actor reads public pages and registers only.