Website Contact Enricher

Turn a list of business websites into public emails, normalized phone numbers, linked social profiles and source evidence. Keep your CRM row IDs, inspect the context around each contact, and see exactly which websites were blocked or only partly scanned.

Use this after collecting company websites from directories, event exhibitor lists or your own CRM. This Actor enriches supplied websites; it does not discover companies, find private employee emails, send messages or verify email delivery. Public beta: test a small batch against your workflow before scaling.

Researching an event? Map Your Show Exhibitor Scraper provides an optional first step and a website-handoff recipe. Each Actor is priced separately.

Quick start

1. Paste your list into Website URLs, one per entry. A homepage, domain or specific branch page works.
2. Optionally add a two-letter Phone country hint, such as US or GB, to normalize national telephone numbers.
3. Keep the three-page default, set your maximum run spend, and select Start.
4. Open Contacts and coverage, then Sources and uncertainty when you need to inspect a value. Export JSON for nested evidence or CSV for the flat email/phone columns.

The prefilled example uses two public business contact pages. Website content and access can change between runs.

{
  "urls": [
    "https://www.pikeplacechowder.com/pikeplacemarketlocation",
    "https://www.pololu.com/contact"
  ],
  "country": "US",
  "maxPages": 3,
  "concurrency": 2,
  "runTimeoutSecs": 60
}

What makes the output useful?

• Evidence for every contact: page URL, extraction method, observation time and nearby source text.
• One row per input: preserve duplicates and IDs rather than silently dropping unsuccessful websites. Completion order can differ; inputIndex restores input order.
• Conservative suggestions: optional primary email and phone fields can stay blank when contacts are ambiguous, special-purpose, conflicted or not supported by the supplied branch hint.
• Purpose hints: distinguish general, sales, support, reservation, press, recruitment, privacy and executive-escalation contexts. These are heuristics, not verified business classifications.
• Phone normalization: equivalent supported formats are consolidated; extensions and link/display disagreements remain available.
• Branch evidence: contacts with explicit structured evidence for other locations are separated into otherLocationEmails and otherLocationPhones. This does not classify every unstructured location block.
• No external API key: no funded email database, SMTP checker, browser-rendering service or residential proxy is required.

Pricing

$0.02 per run start + $0.005 per website where at least one usable public HTML page is inspected ($5 per 1,000). Apify platform usage is included.

A readable website counts once regardless of the number of contacts or pages within your selected limit. Readable websites with no observed contacts and partially readable websites also count. Blocked, failed, invalid and unscanned diagnostic rows have no website-result fee. The start fee still applies, including runs with no readable websites.

Usable websites	Maximum event charge for that run
2	$0.03
10	$0.07
50	$0.27
100	$0.52

These examples assume one run and no pricing discount. The live Pricing tab is authoritative. Use a maximum cost of at least $0.025 for the start and one website. The Actor reserves spending capacity before fetching a website, stops starting work when the allowance is exhausted and saves explicit unscanned diagnostics. Duplicate input entries can each incur a website fee.

Advanced input: preserve CRM IDs or target a branch

Use Advanced website records instead of Website URLs. Clear the URL list before supplying records; sending both lists is rejected to prevent accidental double processing.

{
  "targets": [
    {
      "website": "https://www.pikeplacechowder.com/pikeplacemarketlocation",
      "inputId": "crm-row-42",
      "businessName": "Pike Place Chowder",
      "country": "US",
      "locationHint": "Pike Place Market"
    }
  ],
  "maxPages": 3
}

inputId and businessName are your supplied values, not independently verified identities. country on a record overrides the shared country hint. A location hint is a city or branch label, matched against available source context. Supply the actual branch-page URL when possible. A company-wide or alternate-domain contact may remain in the secondary list without a primary suggestion.

Input	Limits / default
urls or targets	1–100 entries; choose one list
country	Optional uppercase ISO country code
maxPages	1–5 attempted content pages per website; default 3
concurrency	1–2 websites; default 2
runTimeoutSecs	10–600 seconds of work; default 300

The page limit counts attempted content pages. Robots checks and redirects are tracked separately. Each website has a 50-second collection budget. Parsing and saving can extend elapsed time; set the external run timeout at least 60 seconds beyond the work budget. Batches of 100 slow sites may need to be split. Per-page and per-site byte limits bound unusually large responses.

From an exhibitor shortlist to contact evidence

Start with an exported exhibitor shortlist. Select extracted matched profiles, or not_requested when you intentionally used no filters; do not silently treat unknown categories as matches. Review and deduplicate identical website/name pairs while keeping original profile IDs. Different branch URLs or names should remain separate.

Clear Website URLs, then map website to targets[].website, profileId to targets[].inputId, and name to targets[].businessName. Keep each batch at 100 or fewer records. Join output by inputId, because completion order can differ. If you grouped duplicates, preserve the one-to-many mapping outside this Actor. Do not infer a phone country from the event's location.

An owner test on September 30, 2026 UTC reused a completed exhibitor export and sent five matched websites through this Actor. All five IDs returned: four websites yielded usable HTML, two had observed emails and three had phones. One website redirected to another domain and received a free diagnostic row. This is a small technical handoff test, not evidence of paid customer demand or guaranteed email coverage. Inspect any alternate domain before supplying a replacement URL.

The contact stage would cost $0.04 at standard event prices for one start and four readable websites; this is a customer-price illustration, not the developer's own compute bill. Readable websites still count when no contact is observed. The earlier exhibitor stage has separate charges.

Read results correctly

Field	Meaning
emailValues, phoneValues	Flat, deduplicated contact lists for CSV workflows
emails, phones	Contact records with per-value evidence and uncertainty
primaryEmail, primaryPhone	Optional suggestions; not verified contact destinations
primaryEmailStatus	Suggested from source evidence, ambiguous, or not established
emailSourceUrls	Source URLs behind the email list
socials	Profiles linked from the site; ownership unverified
contactForms	Detected forms; never submitted
otherLocationEmails, otherLocationPhones	Explicit other-location structured contacts, where detectable
conflicts	Observed telephone link/display disagreements
pages, robots, metrics	Actual page coverage, access checks and work metrics
billing	Whether the row represents a scanned-website event or a free diagnostic

status is one of:

• complete_within_budget: attempted content pages were usable. This does not mean the whole website was crawled; inspect pendingPages and stopReason.
• partial: at least one usable page, with a failed page or collection limit.
• blocked: access policy, challenge or robots checks prevented collection.
• failed: no usable supported HTML was obtained.
• invalid_input: unsafe or unsupported website URL.
• not_scanned: work or spending limit prevented starting this input.

A blank email list means not observed on the inspected pages, not that the business has no email. deliverability: "not_checked" and reachability: "not_checked" are intentional. Unusual source addresses are preserved with warnings, not silently repaired. A blank primary can coexist with useful secondary contacts.

The Run coverage and work metrics output link opens RUN_SUMMARY, including statuses, readable website count and unscanned input indexes. To continue, start a new run with only the unscanned/unfinished records after reviewing prior output. Resurrect is unsupported to avoid repeat output and charges. Completed rows are saved incrementally if a later operation fails.

Example output

Abbreviated owner-validation result observed September 30, 2026 UTC. Only one of the returned email records is shown. This is a source-observation example, not a deliverability test.

{
  "inputId": "ce-033",
  "website": "https://www.pololu.com/",
  "status": "complete_within_budget",
  "primaryEmail": "inbox@pololu.com",
  "primaryEmailStatus": "suggested_from_source_evidence",
  "emails": [
    {
      "value": "inbox@pololu.com",
      "domainRelationship": "same_domain",
      "evidence": [
        {
          "sourceUrl": "https://www.pololu.com/contact",
          "method": "mailto",
          "raw": "inbox@pololu.com",
          "context": "General questions and comments \u2013 inbox@pololu.com",
          "observedAt": "2026-09-30T01:24:16.775Z",
          "purpose": "general",
          "scope": "unspecified"
        }
      ],
      "warnings": [],
      "deliverability": "not_checked",
      "purposes": [
        "general"
      ],
      "scopes": [
        "unspecified"
      ]
    }
  ]
}

Scope and limitations

This is a static HTTP crawler. It reads supported public HTML, mail/telephone links, visible text, published email obfuscation and selected organization structured data. It follows observed same-host contact/about/location links. It does not execute page JavaScript, guess hidden contact paths, bypass access challenges, submit forms or use a private contact database.

Robots restrictions and unknown robots responses stop collection; sites with crawl delays beyond the budget are reported explicitly. Private-network URLs, credential-bearing URLs, custom ports and unrelated-domain redirects are refused. Use a company's final canonical URL if it moved to a different domain.

Source context, contact purpose and branch association are heuristic. Different websites require different methods; this beta can yield fewer reachable sites than a less restrictive crawler. Phone-format validation is not proof that a number belongs to the business. Use source evidence to decide whether a contact is appropriate for your purpose.

API, automation and AI agents

The API tab provides generated Python, JavaScript and HTTP examples for samstorm/website-contact-enricher. Send the same input JSON, wait for the run to finish, then read its dataset. Check row statuses and RUN_SUMMARY, rather than treating a successful run as success on every website.

You can use Apify's native task, schedule and integration features after confirming your input. For an AI assistant, open this Actor's native MCP setup and select it in Apify's hosted MCP server. Apify authentication and usage charges still apply; no separate data-provider key is needed. An external authenticated MCP client has not been validated for this beta.

Support

Open an issue with the run ID, input website, expected field and a public source URL showing the discrepancy. Include the country or branch hint when relevant. Do not include passwords or API tokens. Reproducible failures and misleading output take priority over expanding the crawl scope.

Changelog

0.2.0 — public beta: simpler URL input, CRM record alternative, source/evidence/diagnostic views, conservative primary suggestions, explicit branch separation, normalized phones, bounded billing and incremental output. Private tests are owner validation, not evidence of paid customer demand or guaranteed accuracy.