Trustpilot Business Scraper - API Based

What this Actor does

Read genuine Trustpilot business summaries without an API key by using the public Trustbox that a business publishes on its own website. The Actor first downloads the supplied company page, reads its published business-unit and template configuration, and requests the same first-party public widget data interface. It checks source robots policies and verifies that the business website and native Trustpilot profile match the publishing company.

This is a website-based business-summary collector. It does not provide global Trustpilot search, scrape excluded profile pages, guess widget templates or retry a business after an entitlement/access denial. The previous keyword-search implementation is replaced by this supported public-widget workflow. The optional query field filters names/domains among the supplied websites; it does not discover new companies.

Input and defaults

The example company page is Bulk Spain's public reviews page: https://www.bulk.com/es/bulk-reviews. You can replace it with one to twenty public company pages that publish a valid Trustpilot widget. Static HTML widget attributes and Adobe Edge Delivery Trustpilot configuration blocks are supported. A company whose widget is injected only through unrelated custom JavaScript may require its server-rendered reviews page.

Example input:

{"startUrls":[{"url":"https://www.bulk.com/es/bulk-reviews"}],"maxItems":10,"minTrustScore":0,"minReviews":0,"sampleMode":false}

Maximum businesses is a cap. One distinct publisher business can produce one record, even when the maximum is ten. Duplicate business IDs are removed. Score, review-count and optional business-name filters can legitimately produce an empty dataset. A source error fails the run rather than adding a diagnostic object to the dataset. No synthetic results or invented contact details are returned.

Output and freshness

Results contain the native display name, business identifier, score, total review count, available star distribution, country, website and exact profile link. Publisher URL, template ID, retrieval timestamp and response hashes provide provenance. Missing email, phone, address or categories are omitted. Regional profile paths are preserved as supplied.

extractedAt is the retrieval time of the first-party public widget response. It is not the time Trustpilot last recalculated a rating. Available source cache-age and date headers are included; their absence is not represented as zero. Only business summaries are emitted, although the native widget response may also contain publicly displayed reviews.

Two anonymized illustrative examples follow; they are documentation examples, not runtime output:

[
 {"businessUnitId":"111111111111111111111111","displayName":"Example Shop","identifyingName":"example.com","trustScore":4.2,"numberOfReviews":125,"trustpilotUrl":"https://www.trustpilot.com/review/example.com","dataMode":"live_public_widget"},
 {"businessUnitId":"222222222222222222222222","displayName":"Example Regional Shop","identifyingName":"example.org/es","trustScore":4.0,"numberOfReviews":40,"trustpilotUrl":"https://es.trustpilot.com/review/example.org/es","dataMode":"live_public_widget"}
]

Failure handling and limits

Only HTTPS company URLs are accepted. Redirects, access denials and missing/invalid widget configuration stop the run without identity, cookie, proxy or template rotation. Requests are sequential, separated per origin, bounded in size and time, and capped at sixty per run. No vendor API key or Apify proxy is required for these source requests. Apify platform execution and configured dataset-event charges still apply in the cloud.

Inspect SOURCE_METADATA and RUN_SUMMARY in the default key-value store. ERROR_DIAGNOSTIC is separate from the dataset on failure. Successful output must be verified locally and on the actual production build before an operator clears maintenance.