Google Maps Lead Intelligence Enricher

Turn existing Google Maps or local-business records into evidence-backed, CRM-ready intelligence. This Actor does not scrape Google Maps and does not judge visual design. It resolves an official web presence, extracts public contacts and factual web signals, filters unsuitable records, and calculates a deterministic opportunity score.

Every conclusion carries a status, confidence, evidence, provenance, reason codes, and an automation-safety decision. A responsive website is only evidence that a site responds; it is not proof that the design is good or bad.

What it does

The Actor normalizes common record aliases, checks supplied websites first, excludes social networks, directories, search engines, and link hubs from owned-site candidates, and follows bounded redirects. When a website is missing, the production default sends two entity-aware query variants through Apify's Google SERP proxy. A query genuinely completed by Google does not call another provider; blocked, failed, or structurally unusable Google responses fall back to public DuckDuckGo HTML and then Bing. It then matches every provider's candidates through the same conservative phone, name, locality, page-content, and schema.org checks.

It writes one dataset row for every input record, including unresolved and per-record error rows. The website states distinguish absence from ambiguity, blocking, and errors. no_owned_website_found is emitted only after a successful search with no owned candidate and no blocking or fetch failure.

Modes are:

• verify: resolve and validate web presence.
• enrich: resolve web presence and extract public contacts and web signals.
• qualify: produce suitability and opportunity decisions. Resolution still runs because web-presence evidence is part of qualification.
• full: all fields plus a short deterministic briefMarkdown.

The output remains top-level schema compatible in every mode. Expensive or unnecessary LLM work is never performed: v1 has no LLM and needs no API key.

Input

Provide either businesses or datasetId. Runtime validation enforces this because the Apify input-schema subset cannot express oneOf.

{
  "businesses": [
    {
      "placeId": "ChIJ-example",
      "title": "Example Veterinary Clinic",
      "streetAddress": "10 High Street",
      "city": "Porto",
      "phoneNumber": "+351 222 333 444",
      "categoryName": "Veterinary clinic",
      "totalScore": 4.7,
      "reviewsCount": 86
    }
  ],
  "mode": "full",
  "maxBusinesses": 100,
  "searchMissingWebsites": true
}

Recognized aliases include name/title/company/companyName, address/streetAddress, phone/phoneNumber/contactPhone, website/url/originalWebsite, mapsUrl/googleMapsUrl, category/categoryName/niche, rating/totalScore, and reviews/reviewsCount/reviewCount.

candidateUrlsByBusiness is an advanced deterministic-pipeline option. Keys are sourceId or normalized source business names and values are arrays of candidate URLs. A matching entry replaces live search for that record and augments any supplied website. It is useful when an upstream search Actor already produced candidates and in reproducible tests:

{
  "businesses": [{ "id": "lead-42", "name": "Example Clinic", "category": "clinic" }],
  "candidateUrlsByBusiness": {
    "lead-42": ["https://example-clinic.test", "https://instagram.com/exampleclinic"]
  }
}

Set includeSourceRecord only when downstream consumers need the original object. includePageText exposes cleaned public text and is off by default. useGoogleSerpProxy defaults to true; disable it to use only the best-effort public path. Google SERP setup is separate from useApifyProxy: the latter and proxyConfiguration only support an optional retry when public DuckDuckGo or Bing is explicitly blocked. The Actor never attempts CAPTCHA evasion, paywall bypass, or private/login-only collection.

Search requests share a 500 ms minimum delay by default, including Google SERP requests, even when businesses run concurrently. Advanced users can set searchDelayMs from 0 to 3000; keep a modest delay for live public search and use zero only for injected tests or controlled environments. A business missing an owned website normally uses two search queries, so the production default uses up to two paid Google SERP requests per such business. At the current reference price of roughly $2.50 per 1,000 SERPs, that is about $0.005 per business; this is approximate and Apify pricing can change.

Chaining from an Apify dataset

{
  "datasetId": "SOURCE_DATASET_ID",
  "mode": "enrich",
  "maxBusinesses": 500,
  "includeSourceRecord": true,
  "useApifyProxy": true,
  "proxyConfiguration": { "useApifyProxy": true }
}

JavaScript API example:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('USERNAME/google-maps-lead-intelligence-enricher').call({
  datasetId: 'SOURCE_DATASET_ID',
  mode: 'full',
  maxBusinesses: 100,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

HTTP API example:

curl -X POST "https://api.apify.com/v2/acts/USERNAME~google-maps-lead-intelligence-enricher/runs?token=$APIFY_TOKEN" \
  -H 'content-type: application/json' \
  -d '{"businesses":[{"name":"Example Clinic","city":"Porto","category":"clinic"}],"mode":"full"}'

Output example

{
  "sourceId": "lead-42",
  "businessName": "Example Clinic",
  "websiteStatus": "owned_website_verified",
  "officialWebsite": "https://example-clinic.test/",
  "confidence": 0.83,
  "safeToAutomate": true,
  "recommendedAction": "monitor",
  "matchedEvidence": [
    { "type": "phone", "value": "+351 222 333 444", "sourceUrl": "https://example-clinic.test/", "weight": 0.55 }
  ],
  "contacts": { "emails": ["hello@example-clinic.test"], "phones": ["+351 222 333 444"], "socialProfiles": [] },
  "businessSuitability": { "status": "suitable", "reasonCodes": [], "score": 1 },
  "opportunity": {
    "score": 45,
    "tier": "medium",
    "components": { "businessStrength": 45, "webPresenceOpportunity": 0, "suitabilityMultiplier": 1 },
    "reasonCodes": ["owned_web_presence_found"],
    "recommendedAction": "consider_outreach"
  }
}

Stable website states are owned_website_verified, owned_website_likely, no_owned_website_found, social_only, directory_only, link_hub_only, dead_owned_domain, ambiguous, blocked, and error. Consumers should automate only when safeToAutomate is true and should still respect businessSuitability.status and recommendedAction.

Pricing and charging

The Actor uses pay-per-event pricing. Each successful output row is charged once according to the selected mode:

Event	Modes	Price
verified-business	verify	$0.020 per business
enriched-business	enrich	$0.025 per business
qualified-business	qualify, full	$0.030 per business

The synthetic apify-actor-start event costs $0.00005 per run for up to 1 GB of memory and covers the first seconds of compute. Error rows are not charged.

Search costs are input-dependent: records with a supplied website can avoid discovery, while a missing owned website normally triggers two Google SERP requests plus any necessary public fallbacks. The event prices already account for normal search, proxy, storage, and compute costs; users are not asked to provide a search API key.

Local development

Requires Node.js 22 or newer.

npm ci
npm test
npm run benchmark
node main.js --test

The unit suite and benchmark use injected HTTP/search fixtures and do not require live network access.

An opt-in live safety observer is also available:

npm run benchmark:live

It runs a small changing public-search observation set without the Apify runtime, labels its summary public_fallback_local, prints DuckDuckGo/Bing provenance and candidate details, and reports how many businesses produced a relevant social, directory, or owned candidate and how many remained ambiguous. Local runs cannot use the in-platform Google SERP proxy. The harness asserts only durable safety invariants. It is intentionally not part of npm test and does not assert that any particular live official domain must be found.

Honest limitations

Public DuckDuckGo and Bing HTML are best-effort fallbacks and can change, rate-limit, challenge, or block automated requests. DuckDuckGo blocking causes a Bing fallback; an explicitly blocked public provider can first retry through the separately configured generic proxy. Google or public proxy use never justifies a negative claim unless both query variants genuinely complete. If Google SERP proxy setup fails, rows carry a stable runWarnings signal and processing continues on the safe public path. Pages rendered entirely by client-side JavaScript may expose little HTML. Phone reuse, similarly named branches, franchise sites, multi-location brands, and missing locality data can remain ambiguous. The chain filter is intentionally narrow to avoid labeling ordinary multi-location businesses as huge chains. Suitability and opportunity are deterministic prioritization aids, not facts about willingness to buy. No technical metadata in this Actor measures visual quality.