Company Email Pattern Finder: find any company's email format from GitHub commits

Find the email format a company uses (first.last@, flast@, first@, firstlast@…) from the public commits of its GitHub organization, with a confidence score, the alternatives, MX records and the mail provider. Every row also brings the company's GitHub profile: website, location, verified badge, public repositories and members, tech stack, top repositories and recent activity.

Give it a list of company domains (stripe.com, nubank.com.br) and get one row per domain. Add the names of people you want to reach and each row also gives their most likely address. It works as an email format lookup and email permutator for B2B outreach, sales prospecting and recruiting research, and as a low-cost alternative to the "email format" lookups of paid email finder tools.

Use cases

• Cold outreach and sales prospecting: know whether a company writes jane.doe@ or jdoe@ before you guess an address.
• Lead enrichment: add email pattern, mail provider (Google Workspace, Microsoft 365…), tech stack and engineering activity to a list of company domains.
• Developer-tool go-to-market: find companies that are active on GitHub, what languages they use and how many people commit.
• Data hygiene: check MX records and spot domains that cannot receive email, without sending a single message.

No SMTP probing, ever. The actor never connects to a mail server and never sends an email: it only reads public commits and DNS MX records. That keeps your IP and domain off blocklists.

What you get per domain

Group	Fields
Email pattern	pattern, patternConfidence (0–0.99), patternConfidenceLevel (high / medium / low / guess), patternExample (the pattern applied to "Jane Doe"), alternatives with their confidence, evidenceCount, candidates for the people you list, roleEmails (jobs@, hello@…), otherCorporateDomains (seen on 2 or more addresses), relatedDomain and its pattern (e.g. mozilla.com for mozilla.org)
Mail	mxValid, mxHosts, mxError, emailProvider (Google Workspace, Microsoft 365, Proofpoint, Mimecast, Zoho…), catchAll (always null: it cannot be known without SMTP)
GitHub organization	githubOrg, orgUrl, orgName, orgDescription, orgBlog, orgEmail, orgLocation, orgTwitter, orgIsVerified, orgFollowers, publicRepos, publicMembers, orgCreatedAt, orgArchivedAt, how sure the match is (orgMatchConfidence, orgMatchReasons)
Activity and stack	lastCommitAt, activeCommitters90d, commitsScanned, scannedRepos, techStack (by bytes of code with a token), topTopics, topRepos (stars, forks, license, homepage, open issues, created and last push dates, default branch), totalStars
Repository inventory	reposFetched, forkRepos, archivedRepos, totalOpenIssues, reposCreatedLast12m (new repositories in the last year), newestRepoCreatedAt, licenses
Counts	uniqueEmailsSeen, corporateEmailsFound, noreplyEmails, personalEmailsSeen, otherDomainEmails
Opt-in	evidence (the corporate emails behind the pattern, with name, repository, date and commit count) with Include evidence emails; personalEmails with Include personal-domain emails

80 fields per domain row, every one present in every row (null when there is no value). Inputs that cannot be processed produce a short error row (row_type: "error", error_type, message) that is never charged.

How well it works (measured)

Measured on 1 October 2026 on 46 company domains chosen before measuring (US, EU and Latin American tech companies), 5 repositories × 300 commits each, 148,821 commits and 8,265 distinct email addresses:

• GitHub organization found: 44 of 46 domains (95.7%), and for 38 of them it is confirmed by commits from the company's own domain.
• Pattern measured from real commits: 38 of 44 companies (86.4%). Confidence 0.75 or higher: 22 of 44 (50%); 0.5 or higher: 28 of 44 (63.6%).
• Latin American companies (n = 8): pattern measured for 5, high confidence for 3.
• Winning patterns: first.last 22, first 7, flast 6, firstlast, firstl and first-last 1 each.
• MX found for 43 of 44 domains; mail providers: Google Workspace 34, Proofpoint 5, Microsoft 365, Mimecast and others 1–2 each.
• Of all addresses seen in commits: 29.2% on the company's domain, 25.1% GitHub noreply, 23.1% personal (gmail, outlook…), 18.5% other companies. Noreply, bots and personal addresses are never used as evidence.

When a company has no GitHub organization, or no commits from its domain, the row says so (notes), gives the MX and provider, and the most common pattern for that provider marked as a guess (patternInferred: true, confidence 0.10).

Pricing

Pay per domain, only for rows delivered:

Event	Price
Company domain with its GitHub organization (pattern, MX, org profile, stack, activity)	$0.03
Company domain without a GitHub organization (MX, provider and a guess)	$0.003
Actor start	$0.00001

What a run costs (measured on Apify, 2 October 2026)

Run	Time	You pay
1 domain (the default input)	16 s	$0.03
3 domains (vercel.com, mercadolibre.com, jetbrains.com)	20 s	$0.09
20 domains, 2 repositories × 100 commits each, no token: GitHub's hourly limit was reached after 17, the other 3 are listed in an error row and not charged	1 min 8 s	$0.51
100 domains	—	$3.00
1,000 domains (with a GitHub token)	—	$30.00

Inputs that cannot be processed (personal email domains, typos) produce an error row that is never charged. A domain is never charged twice in the same run, including after a platform restart. If you set a maximum cost per run, the actor stops cleanly when it is reached.

Input

{
  "domains": ["vercel.com", "mercadolibre.com", "jetbrains.com"],
  "knownPeople": ["Jane Doe, vercel.com"],
  "githubToken": "",
  "maxReposPerOrg": 5,
  "commitsPerRepo": 300,
  "includeSamples": false,
  "includePersonalEmails": false
}

• Company domains: a domain, a website URL or any email of the company. You can also give GitHub organizations directly (stripe or https://github.com/stripe).
• People to build emails for: "Jane Doe" applies to every domain; "Jane Doe, vercel.com" to one.
• GitHub token (optional, recommended): without it GitHub allows 60 API requests per hour, enough for about 15 domains per run; with a token, about 500 domains per hour. A fine-grained token with no permissions, or a classic token with no scopes, is enough (only public data is read). It is never logged or written to the output.
• Where to read commits: Git (default) fetches only commit metadata, about 300 commits per repository in about a second, without using API quota.

Output example

A real row for vercel.com from a run on Apify (2 October 2026, shortened):

{
  "domain": "vercel.com",
  "githubOrg": "vercel",
  "orgName": "Vercel",
  "orgBlog": "https://vercel.com",
  "orgIsVerified": true,
  "publicRepos": 239,
  "publicMembers": 72,
  "orgMatchConfidence": "confirmed",
  "pattern": "first.last",
  "patternConfidence": 0.77,
  "patternConfidenceLevel": "high",
  "patternInferred": false,
  "patternExample": "jane.doe@vercel.com",
  "alternatives": [{ "pattern": "first", "confidence": 0.09, "matches": 2 }, { "pattern": "last", "confidence": 0.03, "matches": 1 }],
  "evidenceCount": 20,
  "mxValid": true,
  "mxHosts": ["aspmx.l.google.com", "alt1.aspmx.l.google.com", "alt2.aspmx.l.google.com"],
  "emailProvider": "Google Workspace",
  "catchAll": null,
  "commitSource": "git",
  "reposScanned": 5,
  "commitsScanned": 1282,
  "uniqueEmailsSeen": 110,
  "corporateEmailsFound": 20,
  "noreplyEmails": 32,
  "personalEmailsSeen": 34,
  "lastCommitAt": "2026-10-02T13:33:47.000Z",
  "activeCommitters90d": 50,
  "techStack": [{ "language": "JavaScript", "repos": 10, "share": 0.468 }, { "language": "TypeScript", "repos": 68, "share": 0.353 }, { "language": "Rust", "repos": 1, "share": 0.151 }],
  "topTopics": ["vercel", "nextjs", "react", "ai", "javascript"],
  "notes": []
}

With Include evidence emails on, evidence lists the addresses behind the pattern, for example {"email": "ana.perez@example.com", "name": "Ana Perez", "patterns": ["first.last"], "kind": "name-match", "source": "commit-author", "repo": "example/app", "date": "2026-09-30T10:00:00Z", "commits": 12}.

How the confidence is computed

Each corporate email in the commits is compared with the commit author's name ("Ana Perez" + ana.perez@ → first.last). A match by name counts 1; an address that only has the shape of a pattern counts 0.3; a named address that matches no pattern counts 0.5 against. Confidence is the pattern's share of the evidence times a saturation by number of matches (1 match 0.5, 2 → 0.75, 3 → 0.88, capped at 0.99), halved when the domain has no MX. Ties are flagged with patternTie.

FAQ

Does it verify that an address exists? No. Verifying needs SMTP probing (RCPT TO), which gets IPs blocklisted and is easy to abuse. The actor gives the pattern, its confidence and the MX records; catchAll is always null.

Why GitHub? Engineers at most tech companies commit with their work address. In the measurement, 29.2% of all addresses in company repositories were on the company's own domain.

What if the company has no GitHub organization? You get the MX, the mail provider and the most common pattern for that provider, clearly marked as a guess, for $0.003.

How is this different from an email finder or verifier? Email finders guess an address and then probe the mail server over SMTP. This actor tells you the company's email naming convention from real evidence, with a confidence score, and never touches the mail server. Pair it with your own verifier if you need per-address validation.

How many domains can I run? Without a token, about 15 per run (GitHub's limit of 60 requests per hour). When the limit is reached the run ends green, delivers what it finished, lists the domains that were not processed and does not charge them. With a token, about 500 per hour.

Is the domain-to-organization match reliable? Each row says how it was matched (orgMatchReasons) and how sure it is (orgMatchConfidence). confirmed means the organization's commits contain addresses on the requested domain (38 of 44 in the measurement).

Do you store personal data? By default the row contains only company-level data: the pattern, counts and the public organization profile. Individual addresses appear only if you turn on Include evidence emails (corporate addresses) or Include personal-domain emails. Turn them on only if you have a lawful basis to process those personal data. GitHub's terms do not allow using its data for spam or for selling personal information.

Can I schedule it or call it from my code? Yes, like any actor: schedules, the API, webhooks and integrations.