Trustpilot Scraper

Amazon is www.amazon.com on Trustpilot, not amazon.com. Every company there sits under the domain Trustpilot picked for it, so the input is a list of those domains, copied from whatever follows /review/ in the company's URL. The wrong form finds nothing at all.

Each review comes back as one row: star rating, title, full review text, author, author country, how many reviews that author has written, verification status, publish and experience dates, language, likes, and the company's reply where there is one.

No login, no API key.

What you get

One row per review. Every row also carries the company summary, so a single review row is enough to know the company's overall score without joining anything:

{
  "companyName": "Amazon",
  "companyDomain": "www.amazon.com",
  "trustScore": 1.6,
  "starRating": 1.5,
  "totalReviews": 48556,
  "categories": [],
  "isClaimed": true,
  "websiteUrl": "https://www.amazon.com",

  "reviewId": "6a87a1e7984a67de41fde1e1",
  "rating": 1,
  "title": "Honestly I love to hate Amazon",
  "text": "Honestly I love to hate Amazon. Naturally I use them because you can find just about anything however the shipping is a joke most times...",
  "author": "Courtnie Hines",
  "authorCountry": "US",
  "authorReviewCount": 2,
  "verified": false,
  "likes": 0,
  "publishedAt": "2026-08-21T02:55:03.000Z",
  "experiencedAt": "2026-08-20T00:00:00.000Z",
  "language": "en",
  "replyText": null,
  "replyAt": null,
  "reviewUrl": "https://www.trustpilot.com/reviews/6a87a1e7984a67de41fde1e1"
}

Two date fields, and they mean different things. publishedAt is when the review was posted. experiencedAt is the date the reviewer says the experience happened, which is the one you want if you're tying complaints to a bad shipping week.

trustScore and starRating also differ on purpose: TrustScore is Trustpilot's weighted figure (recent reviews count for more), starRating is the rounded display value.

Input

{
  "companyDomains": ["www.amazon.com", "nike.com", "booking.com"],
  "maxReviews": 500
}

• companyDomains — the domain as Trustpilot lists it. Open the company's Trustpilot page and copy whatever sits in the URL after /review/. Guessing at it is how you get an empty run.
• companyDomain — a single domain, if one box suits you better than a list.
• maxReviews — per company, not per run. Default 100, ceiling 5,000. Trustpilot serves about 20 per page and the actor paginates.

Optionally push each review into Notion with notionConnector. The dataset gets them regardless.

Limits

• Fields go null. title, text, author, authorCountry, likes, experiencedAt and replyText are all missing sometimes, because Trustpilot omits them. A star-rating-only review with no text is a normal thing on Trustpilot. You aren't charged for empty review shells.
• 5,000 reviews per company per run. For a company with 48,000 reviews you're getting a slice, newest first, not the archive.
• Trustpilot sits behind an AWS WAF challenge. The actor uses rented static addresses that solve it, with rotating datacenter addresses as a fallback. It's reliable in practice, but this is an anti-bot surface and no scraper of it is permanently guaranteed.
• Reviews come back newest first. There's no date-range filter, so set maxReviews and cut the tail yourself.

Proxy

Leave it alone. The default egress is the part that actually works against the WAF. You can supply your own proxy servers and they'll be used exactly as given. Residential groups aren't offered here and a request for one is replaced with datacenter.

Billing

$0.45 per 1,000 reviews, plus a $0.001 run-start fee. Flat on every plan, no volume tiers.

Charged per genuine review delivered. Not charged: empty review shells, blocked look-ups, and companies that returned nothing.