Google Knowledge Panel Scraper

Extract public entity details associated with a Google Knowledge Panel: name, description, type, website, attributes, images, social profiles, related entities, locale, and retrieval provenance. The Actor makes one bounded Google browser request, then a static HTML attempt. When Google is blocked or exposes no usable panel, it can return clearly identified public Wikipedia/Wikidata entity data instead of pretending that fallback data came from Google.

Use cases

• Enrich public organization, person, landmark, and topic records.
• Monitor entity descriptions, official websites, and cited sources.
• Build research datasets with explicit locale and source provenance.
• Export normalized JSON, CSV, Excel, and other Apify dataset formats.

Input

Field	Type	Default	Description
query	string	required	Entity name, 1–300 characters.
gl	string	"us"	Two-letter Google country code.
hl	string	"en"	Language code, such as en or pt-br.
maxRequestRetries	integer	1	Google navigation retries, from 0 to 5.
requestTimeoutSecs	integer	60	Handler timeout, from 15 to 120 seconds.
navigationTimeoutSecs	integer	30	Navigation/static request timeout, from 15 to 120 seconds.
proxyConfiguration	object	direct connection	Optional authorized Apify or custom proxy configuration.

{
  "query": "Albert Einstein",
  "gl": "us",
  "hl": "en",
  "maxRequestRetries": 1,
  "proxyConfiguration": { "useApifyProxy": false }
}

Direct access is the safe default. Google often restricts automated requests; for a cloud run you may select a proxy group already authorized for your Apify account. The Actor never logs or stores proxy credentials.

How retrieval works

The Actor uses these bounded routes in order:

1. A Playwright browser extracts a native Google Knowledge Panel.
2. A static HTTPS request checks for a source-backed panel in Google HTML.
3. Public Wikipedia and Wikidata APIs resolve the entity and merge their complementary fields.

Inspect retrieval.method on every record. nativeGoogleKnowledgePanel and staticGoogleKnowledgePanel are Google-derived. wikipediaSummaryFallback, wikidataEntityFallback, or wikipediaWikidataFallback explicitly identify structured fallback data. Fallback results are useful entity enrichment, but they are not evidence that Google displayed the same fields.

Output

One successful run stores one normalized entity record. Core fields are always present:

Field	Description
entityName	Public entity label.
description	Source-backed entity description.
website	Official, Wikipedia, or Wikidata entity URL selected by source priority.
searchQuery, searchUrl	Original query and canonical HTTPS Google URL.
scrapedAt	ISO 8601 retrieval time.
type, resultType	Stable record identifiers.
searchMetadata, locale	Engine label, query, country, and language.
retrieval	Exact method and public data source.

Optional fields include entityId, entityUrl, subtitle, entityType, attributes, imageUrl, officialWebsite, wikipediaUrl, coordinates, aliases, socialProfiles, relatedEntities, and sourceUrls. Missing optional values are omitted rather than replaced with fabricated placeholders.

{
  "entityName": "Albert Einstein",
  "description": "German-born theoretical physicist",
  "website": "https://en.wikipedia.org/wiki/Albert_Einstein",
  "searchQuery": "Albert Einstein",
  "scrapedAt": "2026-08-30T10:00:00.000Z",
  "type": "knowledge-panel",
  "resultType": "knowledge-graph",
  "searchUrl": "https://www.google.com/search?q=Albert%20Einstein&hl=en&gl=us",
  "searchMetadata": {
    "engine": "google-knowledge-panel",
    "country": "US",
    "language": "en"
  },
  "locale": { "country": "US", "language": "en" },
  "retrieval": {
    "method": "wikipediaWikidataFallback",
    "source": "Wikipedia and Wikidata public entity data"
  }
}

Cost and performance

Browser startup is the main compute cost. Concurrency is fixed at one because every run contains one entity query. Lower retry and timeout values reduce blocked-target spending. Wikipedia/Wikidata requests are small and bounded; the Actor deduplicates merged profiles, related entities, aliases, and source URLs.

Limitations and responsible use

• Google layout and panel availability vary by query, locale, IP reputation, and time.
• A fallback record may differ from a current Google panel; rely on retrieval when interpreting results.
• Search can resolve an ambiguous name to the most prominent public entity. Use a specific query when names collide.
• Respect source terms, copyright, privacy, and applicable law. Do not use public entity data to infer sensitive personal traits.

FAQ

Why did I receive Wikipedia/Wikidata data? Google was blocked or had no complete panel, so the Actor used its public structured fallback and labeled the method explicitly.

Why did the run fail instead of returning an empty object? No route produced the required public name, description, and website. Failing transparently prevents misleading placeholder records.

For support, provide the run ID, sanitized input, locale, and first error message. Never include proxy credentials or private data.