Trustpilot Reviews: Deep Scrape with Star Filter

For each Trustpilot business slug, scrape reviews via the page's __NEXT_DATA__ JSON. One row per review. Optional star-filter pagination loops the business with stars=1..5 to multiply effective depth past Trustpilot's standard per-business window.

What you get

{
  "businessSlug":  "apify.com",
  "reviewId":      "abc123",
  "stars":         5,
  "title":         "Great service",
  "text":          "Used Apify for 3 months, saved us 40 hours weekly.",
  "language":      "en",
  "dateCreated":   "2026-05-12T08:30:00Z",
  "dateExperience":"2026-05-10",
  "reviewer": {
    "country":     "US",
    "numReviews":  3
  },
  "ownerReply": {
    "text":        "Thanks for the kind words, happy to hear it!",
    "date":        "2026-05-13T10:00:00Z"
  },
  "sentimentTags": ["positive_signal", "product_praise"]
}

Reviewer names and ids are not included; e-mail addresses and phone numbers inside review and reply text are replaced with [email removed] / [phone removed].

How to use

Input. A list of business slugs (apify.com) or full URLs (https://www.trustpilot.com/review/apify.com). URLs are reduced to slug.

Star-filter pagination. Trustpilot's default review feed exposes a limited number of pages per business (the actor reads up to 24 pages per feed). Turning on useStarFilterPagination runs the same business with stars=1, stars=2, …, stars=5 filters one after another, which reaches more reviews than the single feed. The total per business is capped by maxReviewsPerBusiness (default 200).

v0.2: what's new

sentimentTags. Every review is classified against an 8-tag lexicon based on Trustpilot's idiom. Tags are independent (a review can have several or none):

tag	what it signals
positive_signal	"excellent", "love it", "highly recommend", …
negative_signal	"terrible", "worst", "disappointed", …
scam_signal	"scam", "fraud", "ripped off", "stole my money", …
shipping_complaint	"shipping", "delivery", "arrived late", …
support_complaint	"no response", "ignored", "customer service", …
refund_complaint	"refund", "money back", "chargeback", …
price_complaint	"overpriced", "not worth", "hidden fee", …
product_praise	"quality", "well made", "works perfectly", …

Filter for scam_signal to surface fraud reports across a portfolio; filter for shipping_complaint over a 30-day window to track logistics regression.

Delta mode. Set deltaMode: true to charge only for new reviews since the last run. The actor persists a per-business seen-ID set in its KV store. First run emits everything; subsequent runs skip rows whose reviewId was already seen. Cap: 25 000 ids per business.

Use case, daily reputation monitoring: schedule a daily run with deltaMode: true, pay $0.0003 per new review only, get an empty dataset on quiet days.

Pricing

PAY_PER_EVENT · $0.0003 per review_extracted · 3 000 reviews ≈ $1.

With delta mode on, you only pay for reviews you haven't seen before.