BE-Lead-Hunter

An experimental business-contact research and evidence-review Actor. It helps organize supplied business listings, extract contacts from saved website HTML, and distinguish supported observations from checks that have not been performed.

This is a developing research tool, not an automated sales or outreach platform. It does not start discovery Actors, verify mailboxes, assess buying intent, or send messages. An extracted public email is not permission to contact someone.

What it can do

Run mode	Purpose and limits
snapshots	Review supplied listings and saved HTML without fetching websites.
discovery_plan	Produce an offline discovery plan; does not execute it.
reuse_discovery	Read a compatible completed Google Maps Scraper run when explicitly enabled and permitted; starts no new source run.
crawl_simulation	Exercise collection using supplied synthetic HTTP/DNS fixtures.
crawl_websites	Bounded public-website collection; requires both input opt-in and server authorization. Robots-policy limitations can cause collection holds.
collect_discovery	Website collection for a developer-reviewed source/profile, not arbitrary unreviewed source batches.
qualify_saved_collection	Checksum-pinned replay of a developer-reviewed saved collection. Not available for arbitrary new batches.
save_reviews	Separately authorized append-only review persistence. Not a generally configured hosted database service.

For a first test, use snapshots with fictional supplied data. Leave all live, source-read and persistence opt-ins off. Never put credentials in Actor input.

Minimal fictional example

{
  "runMode": "snapshots",
  "country": "United States",
  "state": "Texas",
  "city": "Demo City",
  "category": "real estate",
  "enableLiveRequests": false,
  "allowReadOnlyApify": false,
  "enableReviewPersistence": false,
  "discoveryItems": [
    {
      "placeId": "fictional-cedar-realty",
      "title": "Fictional Cedar Realty",
      "address": "100 Sample Street, Demo City, TX",
      "website": "https://cedar-realty.example/"
    }
  ],
  "websitePages": [
    {
      "url": "https://cedar-realty.example/",
      "statusCode": 200,
      "observedAt": "2026-10-20T00:00:00Z",
      "html": "<html><head><title>Fictional Cedar Realty</title></head><body><h1>Fictional sample</h1><p>This fictional business demonstrates saved-page extraction and research review. It is not a real contact or live website audit.</p><a href=\"mailto:hello@cedar-realty.example\">Contact</a></body></html>"
    }
  ]
}

The example supplies HTML as data; no request is made to the .example domain.

Output

The dataset contains review records with identity, location, contact provenance, evidence limitations and disabled sending. Some modes also separate deferred or rejected records in key-value-store reports. Saved-collection qualification retains all source records in its dataset, including holds and rejects.

RUN_SUMMARY describes the executed mode and counts. Saved-collection qualification also writes QUALIFICATION_REPORT, including its input-integrity checks. Records require human review; an evidence-readiness score is not lead quality, mailbox deliverability, service need or purchase intent.

Fourteen service categories are represented in saved-collection qualification, but most checks are not implemented. Current evidence checks are limited to basic page title/description/heading and viewport signals. Missing evidence is unknown, not proof that a business needs a service. This is not a comprehensive SEO, accessibility, performance, advertising or social-media audit.

Costs, access and responsible use

Apify compute, storage and transfer can incur usage charges, including on runs that perform no new website scraping. Input budget declarations are not billing caps; set a small Console/API run limit and inspect usage. Optional source tools have their own pricing and terms; this Actor does not start them for you.

Use only data and websites you are authorized to process. Keep private inputs, credentials and real customer/lead exports out of public examples and share links. Sending, mailbox verification, AI analysis, n8n activation and production database connectivity are not part of the validated saved-evidence qualification workflow.

Validation status

The saved-evidence integration passed 488 cloud tests and reproduced the reviewed ten-record pilot exactly in exported report, summary and dataset comparisons. That pilot has partial website coverage and no outreach-qualified records. Broader markets, new source profiles and production reliability require separate validation. The tool remains experimental.