Lead Enricher — Websites & Facebook Pages → Emails, Phones, Socials

Give it any list of business URLs — websites and Facebook Pages, mixed — and get back one clean, unified contact record per URL: emails, phones, social links, website, and (for Pages) followers and intro. Each URL is auto-routed to the right extractor, so you don't have to sort your list first.

Most enrichment tools make you pick a source and pay per credit. This takes a raw, mixed list of URLs and returns deduped contacts in one pass — ideal for turning a scraped org/venue/agency/prospect list into an outreach-ready sheet.

What it does

• Auto-routes every URL. A facebook.com/... Page → the Facebook extractor; any other website/domain → the website contact extractor. You just paste URLs.
• Unified output. One consistent record per input URL — no reconciling two different formats.
• Deep website scan. For websites, optionally also checks /contact, /contact-us, and /about for a much higher hit rate.
• Emails, phones, socials — plus website + follower count + intro for Facebook Pages.

Output (one row per URL)

input_url, source (website / facebook), website, emails[], phones[], socials{}, other_urls[], name, followers, intro, scrapedAt, provider_status.

URLs are trimmed, blank entries removed, and exact duplicates collapsed. Output contains exactly one row per remaining URL, in input order, including when providers fail or return no usable item. Empty contacts are arrays, missing scalar values are null, and followers may be a number or a string such as "408K".

provider_status contains website, facebook_primary, and facebook_fallback, each with status and error:

Status	Meaning	Error code
success	A matching item was processed; contacts may still be empty	null
no_item	No attributable item was returned	NO_ITEM
malformed	A matching item had invalid field types; valid fields were retained	MALFORMED_ITEM
error	Call/run failure, dataset read failure, malformed dataset envelope, or item-reported failure	CALL_FAILED, DATASET_FAILED, MALFORMED_DATASET, or PROVIDER_ERROR
not_run	Provider was inapplicable, disabled, or unnecessary	null

Unattributable malformed items and unexpected URLs are ignored; they never create extra rows. Matching uses normalized URLs, retaining Facebook profile IDs. Providers must return the original URL in url (or pageUrl / facebookUrl for Facebook); redirects without an attributable original URL remain no_item. Duplicate items merge contacts. Valid data survives malformed fields, subsequent dataset failures, and fallback failures. Failure evidence remains visible even if another item succeeds.

The optional Facebook fallback runs only for rows without an email, phone, or website. It adds contacts and fills missing metadata without replacing existing successful fields. Both provider outcomes remain visible. Raw provider errors and exceptions are never copied into records or application logs; only fixed error codes are exposed.

Use cases

• Enrich a scraped lead list — feed organizer/venue/agency/company URLs and get contactable rows for your CRM or outreach.
• Mixed sources, one run — some prospects only have a website, others only a Facebook Page; handle both without splitting the list.
• Agent / MCP workflows — a single "URL → contacts" tool an AI agent can call to enrich a company on the fly.

Input

• urls (required) — website and/or Facebook Page URLs, mixed.
• crawlContactPages — also scan /contact and /about on websites (recommended).
• facebookFallbackActorId — optional alternate actor for Facebook rows still lacking contact data; disabled by default.
• proxyConfiguration — optional, passed to the Facebook extractor.

Pricing

Rows with an email or phone use the enriched-lead billing event. Rows without these contacts use basic-lead only when a provider succeeded. Failed or missing placeholders are not billed as basic leads. Underlying subactor usage may still incur platform charges.

Support

Questions or a custom enrichment workflow? business@aitorsm.com

Local validation

Run npm install, then npm run build, npm run typecheck, npm test, and git diff --check. Tests use a mocked Actor client and never call external actors. Do not use npm start for offline testing: it runs the real orchestration.