Trustpilot Reviews Scraper & Sentiment Analytics – Reviews, TrustScore, Company Data

Trustpilot Reviews Scraper & Sentiment Analytics extracts public Trustpilot data without a Trustpilot account or API key: customer reviews, company profiles (TrustScore, contacts, categories), keyword search results, and category listings as clean JSON, CSV, or Excel. Turn on review sentiment analysis to get an investor-grade insight report per company: rating confidence interval, 90-day trend, complaint and praise themes with evidence quotes, reply behaviour, and review-integrity flags. No LLM and no extra API keys.

In one sentence: give it a company domain (apify.com), a keyword (insurance agency in United States), or a category (car_dealer) and get structured rows tagged business, review, search_result, insight, or benchmark.

---

Quick start (60 seconds)

1. Open the Actor and choose a Mode.
2. Add companyUrls, searchQueries, or categories.
3. (Optional) Tick Analyze reviews for sentiment and insight rows.
4. Click Start and download the dataset as JSON, CSV, Excel, or XML, or read it through the API.

{
  "mode": "companyReviews",
  "companyUrls": ["apify.com"],
  "maxReviews": 100,
  "languages": "en",
  "sort": "recency"
}

Add sentiment and insight analysis:

{
  "mode": "companyReviews",
  "companyUrls": ["apify.com"],
  "maxReviews": 0,
  "sort": "recency",
  "analyzeReviews": true
}

---

What can I scrape from Trustpilot?

Mode	Best for	Input	Output row types
companyReviews	Reviews and TrustScore for known companies	companyUrls	business (profile) + review (+ insight with analysis)
search	Finding companies by keyword or niche	searchQueries, country	search_result (+ review and insight if scrapeReviews is on)
category	Listing companies in a Trustpilot category	categories	search_result (+ review and insight if scrapeReviews is on)

Data you get

• Reviews: title, full text, 1–5 star rating, published and experience dates, language, helpful votes, verification level and source, reviewer name/country/review count, company reply text and date. With analysis on, each review also gets sentiment, sentimentScore, and textSentimentScore.
• Company profiles: TrustScore, total reviews, reviews in the last 12 months, claimed/closed status, website, categories and breadcrumbs, address, city, country, and email/phone/websites when Trustpilot publishes them.
• Search & category hits: name, slug, TrustScore, review count, categories, location, contacts when shown, and profile URL.
• Insight rows (analysis on): one per company with ratings, trend, themes, responsiveness, integrity flags, red/green signals, and a plain-English executive summary.
• Benchmark row (analysis on, 2+ companies): companies ranked against each other on shared metrics.

---

Why use this Trustpilot scraper?

• No Trustpilot login or API key: works from public profile, search, and category pages.
• Three modes in one Actor: reviews, discovery, and category mining without juggling tools.
• Built-in sentiment and insight analytics: statistics, not just a score. Every theme links back to real review quotes, so analysts can audit it.
• Honest about sample limits: trend verdicts say insufficient_data when the sample cannot support them, and filtered runs are flagged automatically.
• Review-integrity signals: invited-vs-organic rating gap, rating bursts, duplicate text, and first-time reviewer share.
• Server-side review filters: stars, date range, language, topics, verified only, with replies only, in-review keyword, and sort order.
• Discovery to insight in one run: scrapeReviews: true plus analyzeReviews: true analyzes every company found by search or category.
• Typed, flat rows: every item has a type field, so there are no nested surprises for BI tools or agents.
• Automation-ready: Apify API, schedules, webhooks, and integrations (Make, n8n, Zapier, Google Sheets), plus MCP for AI agents.

---

Use cases

• Investor and VC due diligence: check whether a company's headline rating is organic, stable, and well managed before you rely on it.
• Reputation and brand monitoring: track new reviews, TrustScore, and rising complaint themes on a schedule.
• Voice of Customer (VoC) and sentiment analysis: feed review text and sentiment fields to dashboards or NLP pipelines.
• Competitor benchmarking: compare rating, net star score, reply behaviour, and integrity signals across rivals in one run.
• Review authenticity checks: spot invitation-driven rating inflation, one-day 5-star bursts, and duplicated text.
• Churn and support insights: find churn language, refund friction, and slow-support themes in negative reviews.
• Lead generation: find companies in a niche or country and capture published contact details.
• AI agents and RAG: give an agent live, structured customer-sentiment data on demand.

---

How do I scrape Trustpilot reviews for a company?

Use companyReviews mode and pass a domain or a Trustpilot profile URL. All of these are valid:

• apify.com
• www.adorama.com
• https://www.trustpilot.com/review/webuyanycarusa.com

Set maxReviews to the number of reviews you want per company. Use 0 for as many as public pages allow (Trustpilot limits how deep pagination goes, so very large review sets may be only partly available).

How do I find companies on Trustpilot by keyword?

Use search mode with searchQueries (for example "Travel agency in New York") and a 2-letter country. Add scrapeReviews: true to also collect reviews for each company found.

How do I scrape a whole Trustpilot category?

Use category mode with category slugs (car_dealer, technical_service, insurance_agency) or full URLs such as https://www.trustpilot.com/categories/car_dealer.

---

How does Trustpilot review sentiment analysis work?

Set analyzeReviews: true. The Actor then does two things:

1. Enriches every review row with sentiment (positive, neutral, negative), sentimentScore (−1 to 1), and textSentimentScore.
2. Pushes one type: "insight" row per company, summarizing the whole review sample. With 2+ companies it also pushes one type: "benchmark" row.

How sentiment is scored

• The star rating is mapped to −1…+1 (1★ = −1, 3★ = 0, 5★ = +1).
• English review text is scored with VADER, a rule-based sentiment model. It is not an LLM.
• The final sentimentScore blends both 50/50. Labels: at or above +0.10 is positive, at or below −0.10 is negative, otherwise neutral.
• If the text is not English, textSentimentScore is null and the score uses the star rating alone.

What an insight row tells you

Section	What it answers
sample	How many reviews were analyzed, coverage of the profile, confidence (low / medium / high), and warnings you should read first
ratings	Average rating with 95% confidence interval, star distribution, net star score, polarization, TrustScore
sentimentSummary	Share of positive / neutral / negative reviews and average sentiment
momentum	90-day rating trend and review-volume trend, plus monthly and quarterly series
responsiveness	Reply rate, reply rate to 1–2 star reviews, median and 90th-percentile reply time
themes	Complaint, praise, and latent friction themes with prevalence, trend status, and evidence quotes
distinctiveTerms	Words and phrases that statistically separate negative from positive reviews
integrity	Invited share, invited-vs-organic rating gap, bursts, duplicate text, first-time reviewers
audience	Reviewer countries, languages, and business/team/developer usage signals
signals	Red and green signals, each with a code, severity, and detail
keyTakeaways, executiveSummary	Plain-English summary for people who won't read the JSON

Key metrics explained

• Net star score: share of 4–5 star reviews minus share of 1–2 star reviews, from −100 to +100.
• Polarization: share of reviews that are exactly 1 or 5 stars. High values mean a split customer base that an average hides.
• Rating trend (momentum.ratingTrend): compares the last 90 days with the prior 90. A verdict of improving or declining needs at least 20 reviews in each window, a change of at least 0.15 stars, and statistical significance (Welch t-test). Otherwise you get stable or insufficient_data, with a reason.
• Volume trend (momentum.volume): accelerating, stable, or slowing. Burst days are excluded when they fall in the windows, because invitation campaigns and brigading would otherwise distort the ratio. The raw counts stay visible.
• Themes: matched on whole words and phrases (so "no bugs" or "never had an outage" are not counted as complaints). Each theme shows its share of the group, a trend status (rising, stable, fading, or insufficient_data), and up to two evidence quotes with reviewId, rating, date, and snippet. Reviewer names are not included in quotes.
• Latent friction: complaint themes found inside 4–5 star reviews. Even happy customers can flag pricing, setup, or limits, and this is where it shows up.
• Small samples: each theme reports groupSize and lowSample. High-severity flags use a confidence bound, so 13 fraud mentions among 18 negative reviews can raise a flag while 1 mention among 18 cannot.

Review-integrity signals

These are heuristics for a human to investigate, not proof of manipulation.

Signal	What triggers it
reviews_mostly_invited	80% or more of reviews were collected through company invitations
invited_rating_inflation	Invitations make up at least half the sample and invited reviews rate at least 0.5 stars higher than organic ones
coordinated_praise_burst / coordinated_attack_burst	A day with an unusual spike of reviews that are 80% or more 5-star (or 1-star)
duplicate_review_text	3% or more of reviews share identical text
mostly_first_time_reviewers	60% or more of reviewers have only ever written one review (common with invitations)
many_contentless_5_star	30% or more of 5-star reviews are four words or fewer
rating_text_mismatch	5% or more of reviews have a star rating contradicting their text

Other red signals: rating_declining, rising_complaint, fraud_allegations, churn_or_warning_language, ignores_negative_reviews, slow_replies_to_negatives, polarized_ratings. Green signals: rating_improving, strong_net_star_score, responsive_to_negatives, review_volume_accelerating.

Get reliable results

• Set maxReviews to 0, or at least 500 for large companies. A small sample cannot support trend claims.
• Sort by recency and do not use stars, topics, verifiedOnly, withReplies, or reviewSearch. These filters change which reviews exist in the sample. When any is set, the insight row is marked filtered, confidence drops to low, trend verdicts are disabled, and only integrity signals are kept.
• Read sample.warnings and sample.confidence first. They list the limits of that run, such as low coverage, few negative reviews, or sentiment running on ratings only.
• Use English reviews for text sentiment. Other languages are scored from the star rating alone.

---

Input parameters

Parameter	Type	Default	Applies to	Description
mode	string	companyReviews	all (required)	companyReviews, search, or category
companyUrls	string[]	–	company	Domains or profile URLs
searchQueries	string[]	–	search	Keywords or business descriptions
categories	string[]	–	category	Category slugs or URLs
maxReviews	integer	100	company; search/category with scrapeReviews	Max reviews per company (0–50,000). 0 = all available public pages. Use 0 or 500+ with analyzeReviews
maxResults	integer	50	search, category	Max business rows (0–10,000). 0 = unlimited within maxPages
maxPages	integer	5	search, category	Pagination depth (1–50)
country	string	US	search	2-letter country code for search ranking (US, GB, DE)
scrapeReviews	boolean	false	search, category	Also scrape reviews for each discovered company
analyzeReviews	boolean	false	all modes that scrape reviews	Add sentiment fields to reviews, push one insight row per company, and a benchmark row for 2+ companies
stars	string[]	all	reviews	Only these ratings: "1" to "5"
date	string	all	reviews	all, last30days, last3months, last6months, last12months
languages	string	en	reviews	ISO language code (en, es, de)
topics	string[]	–	reviews	Trustpilot topic IDs (see below)
verifiedOnly	boolean	false	reviews	Keep only verified reviews
withReplies	boolean	false	reviews	Keep only reviews with a company reply
reviewSearch	string	""	reviews	Keyword filter inside review text
sort	string	recency	reviews	recency, relevance, or rating
includeBusinessProfile	boolean	true	reviews	Emit one business row before each company's reviews

Note: scrapeReviews defaults to false, so search and category runs return company listings only. Set it to true to also collect reviews for every company found (uses maxReviews and the review filters).

Topic IDs: price, service, staff, value_for_money, customer_service, recommendation, location, ethics, booking_process, payment, credibility, website, user_experience, delivery_service

---

Input examples

Recent verified 1–2 star reviews for several companies

{
  "mode": "companyReviews",
  "companyUrls": [
    "www.adorama.com",
    "https://www.trustpilot.com/review/webuyanycarusa.com"
  ],
  "maxReviews": 200,
  "stars": ["1", "2"],
  "date": "last12months",
  "languages": "en",
  "verifiedOnly": true,
  "sort": "recency"
}

Reviews mentioning a keyword, with company replies only

{
  "mode": "companyReviews",
  "companyUrls": ["apify.com"],
  "reviewSearch": "support",
  "withReplies": true,
  "maxReviews": 50
}

Investor-style insight on one company

{
  "mode": "companyReviews",
  "companyUrls": ["apify.com"],
  "maxReviews": 0,
  "sort": "recency",
  "languages": "en",
  "analyzeReviews": true
}

Competitor benchmark (insight rows plus a benchmark row)

{
  "mode": "companyReviews",
  "companyUrls": ["www.adorama.com", "webuyanycarusa.com", "apify.com"],
  "maxReviews": 0,
  "sort": "recency",
  "analyzeReviews": true
}

Company discovery only (no reviews)

{
  "mode": "search",
  "searchQueries": ["Insurance agency in United States"],
  "country": "US",
  "maxResults": 40,
  "maxPages": 2,
  "scrapeReviews": false
}

Discovery plus reviews and insight in one run

{
  "mode": "search",
  "searchQueries": ["Travel agency in New York"],
  "country": "US",
  "maxResults": 10,
  "maxPages": 1,
  "scrapeReviews": true,
  "maxReviews": 200,
  "sort": "recency",
  "analyzeReviews": true
}

Category listings

{
  "mode": "category",
  "categories": ["technical_service", "car_dealer"],
  "maxResults": 50,
  "maxPages": 3,
  "scrapeReviews": false
}

---

Output

Every dataset item has a type field. Filter on it in the Apify console, API, or your code.

type	Meaning	Produced by
business	Full company profile	companyReviews (when includeBusinessProfile is true)
review	One customer review	all modes (company reviews, or discovery with scrapeReviews)
search_result	Lighter company card from a listing	search, category
insight	Per-company sentiment and investor analytics	any run with analyzeReviews: true
benchmark	Peer ranking across companies	analyzeReviews: true with 2+ companies

Example: business row

{
  "type": "business",
  "businessUnitId": "62ac9c3177fbadc00785d7fe",
  "displayName": "Apify",
  "identifyingName": "apify.com",
  "trustScore": 4.8,
  "stars": 5,
  "numberOfReviews": 811,
  "numberOfReviewsLast12Months": 536,
  "websiteUrl": "https://apify.com",
  "profileUrl": "https://www.trustpilot.com/review/apify.com",
  "countryCode": "CZ",
  "isClaimed": true,
  "isClosed": false,
  "isTemporarilyClosed": false,
  "isCollectingReviews": false,
  "categoryId": "software_company",
  "categoryName": "Software Company",
  "categories": [
    { "id": "software_company", "name": "Software Company", "isPrimary": true, "rank": null },
    { "id": "software_vendor", "name": "Software Vendor", "isPrimary": false, "rank": null }
  ],
  "breadcrumbTop": "Electronics & Technology",
  "breadcrumbMid": "Internet & Software",
  "breadcrumbBottom": "Software Company",
  "address": "Vodičkova 704/36, Nové Město",
  "city": "Prague",
  "zipCode": "110 00",
  "country": "CZ",
  "email": "hello@apify.com",
  "emails": ["hello@apify.com"],
  "phone": "",
  "phones": [],
  "websites": ["https://apify.com"],
  "verifiedPaymentMethod": false,
  "verifiedByGoogle": false,
  "verifiedUserIdentity": false,
  "scrapedAt": "2026-10-23T08:56:10.853755+00:00"
}

(Category list shortened for readability.)

Example: review row (with analysis on)

{
  "type": "review",
  "reviewId": "6a4e86e661133d86dacdfa28",
  "businessUnitId": "62ac9c3177fbadc00785d7fe",
  "businessName": "Apify",
  "identifyingName": "apify.com",
  "profileUrl": "https://www.trustpilot.com/review/apify.com",
  "title": "Helped me a lot with scraping data…",
  "text": "Helped me a lot with scraping data reliably with ai for my market intelligence dashboards.",
  "rating": 5,
  "likes": 0,
  "language": "en",
  "source": "BasicLink",
  "isVerified": false,
  "verificationLevel": "invited",
  "verificationSource": "invitation",
  "publishedDate": "2026-07-08T19:20:38.000Z",
  "experiencedDate": "2026-07-08T00:00:00.000Z",
  "consumerName": "Example Reviewer",
  "consumerCountry": "HU",
  "consumerReviewsCount": 2,
  "hasReply": false,
  "replyMessage": "",
  "replyPublishedDate": "",
  "isPending": false,
  "filtered": false,
  "scrapedAt": "2026-10-23T08:57:04.821191+00:00",
  "textSentimentScore": 0.4767,
  "sentimentScore": 0.7384,
  "sentiment": "positive"
}

(Reviewer name replaced and text lightly edited for privacy. The three sentiment fields appear only when analyzeReviews is on.)

Example: insight row

An illustrative insight row for an anonymized company, trimmed for readability. Values are rounded.

{
  "type": "insight",
  "schemaVersion": 3,
  "identifyingName": "example.com",
  "displayName": "Example Co",
  "asOf": "2026-10-23T09:36:01+00:00",
  "sample": {
    "reviewCountAnalyzed": 100,
    "profileReviewCount": 108,
    "coverage": 0.9259,
    "oldestReview": "2025-11-14T11:48:04+00:00",
    "newestReview": "2026-10-22T09:52:44+00:00",
    "reachesPriorWindow": true,
    "englishTextShare": 1,
    "confidence": "medium",
    "filtered": false,
    "scope": { "filtersApplied": [], "date": "all", "language": "all", "sort": "recency" },
    "sentimentEngine": "vader",
    "warnings": ["Only 18 negative reviews in the sample; complaint themes are anecdotal."]
  },
  "ratings": {
    "avgRating": 4.23,
    "ci95": [3.93, 4.53],
    "distribution": { "5": 0.78, "4": 0.02, "3": 0.02, "2": 0.01, "1": 0.17 },
    "netStarScore": 62,
    "polarization": 0.95,
    "trustScore": 4.2
  },
  "sentimentSummary": {
    "distribution": { "positive": 0.8, "neutral": 0.03, "negative": 0.17 },
    "avgScore": 0.5389
  },
  "momentum": {
    "ratingTrend": {
      "recent90d": { "reviews": 19, "avgRating": 4.316 },
      "prior90d": { "reviews": 56, "avgRating": 4.286 },
      "delta": null,
      "tStat": null,
      "verdict": "insufficient_data",
      "reason": "fewer_than_20_reviews_in_a_window"
    },
    "volume": {
      "recent90d": 19,
      "prior90d": 56,
      "recent90dExBursts": 19,
      "prior90dExBursts": 28,
      "burstAdjusted": true,
      "ratio": 0.68,
      "verdict": "slowing"
    },
    "monthly": [
      { "month": "2026-09", "reviews": 2, "avgRating": 3.5 },
      { "month": "2026-10", "reviews": 17, "avgRating": 4.412 }
    ]
  },
  "responsiveness": {
    "replyRate": 0.17,
    "negativeReviews": 18,
    "negativeReplyRate": 0.3333,
    "medianReplyHours": 111.8,
    "p90ReplyHours": 977,
    "negativeMedianReplyHours": 476
  },
  "themes": {
    "complaints": [
      {
        "theme": "refund / money-back friction",
        "mentions": 4,
        "groupSize": 18,
        "lowSample": false,
        "shareOfGroup": 0.2222,
        "status": "insufficient_data",
        "evidence": [
          {
            "reviewId": "6a00000000000000000000aa",
            "rating": 1,
            "publishedDate": "2026-08-14T10:02:11+00:00",
            "quote": "…asked for a refund three times and nobody answered…"
          }
        ]
      }
    ],
    "praise": [
      { "theme": "easy to use", "mentions": 23, "groupSize": 80, "shareOfGroup": 0.2875, "status": "stable" }
    ],
    "latentFriction": []
  },
  "integrity": {
    "invitedShare": 0.54,
    "sourceMix": [
      { "value": "BasicLink", "share": 0.54 },
      { "value": "Organic", "share": 0.46 }
    ],
    "invitedVsOrganicRatingGap": 1.513,
    "firstTimeReviewerShare": 0.79,
    "duplicateTextShare": 0,
    "burstDays": [
      { "date": "2026-06-18", "reviews": 18, "avgRating": 4.94, "share5Star": 0.944, "share1Star": 0 }
    ]
  },
  "signals": {
    "red": [
      { "code": "coordinated_praise_burst", "severity": "high", "detail": "18 reviews on 2026-06-18, overwhelmingly 5-star." },
      { "code": "invited_rating_inflation", "severity": "medium", "detail": "Invited reviews average +1.51 stars vs organic; headline rating may be flattered." }
    ],
    "green": [
      { "code": "strong_net_star_score", "severity": "medium", "detail": "Net star score +62 (share 4-5 star minus share 1-2 star)." }
    ]
  },
  "executiveSummary": "Example Co: 100 reviews analyzed (93% of 108 on profile), confidence medium. Average rating 4.23/5 (95% CI 3.93-4.53), net star score +62. Profile TrustScore 4.2. Rating trend: insufficient data; review volume: slowing. Replies to 33% of negative reviews. 4 red flag(s), 1 positive signal(s). Caveat: Only 18 negative reviews in the sample; complaint themes are anecdotal. Main concern: 18 reviews on 2026-06-18, overwhelmingly 5-star."
}

The real row also contains distinctiveTerms, topicShares, audience, keyTakeaways, quarterly, and a few legacy summary fields (reviewCountAnalyzed, sentimentDistribution, avgSentimentScore, ratingDistribution, verifiedShare, replyRate, avgReplyHours, topTopics, risingComplaints, topPraise, trendLast90Days) kept for backward compatibility.

Example: benchmark row (shape)

{
  "type": "benchmark",
  "schemaVersion": 3,
  "peerCount": 3,
  "note": "Percentiles are relative to this peer set only (100 = best). Low-confidence companies are ranked but should be read cautiously.",
  "companies": [
    {
      "identifyingName": "example-a.com",
      "displayName": "Example A",
      "reviewCountAnalyzed": 480,
      "confidence": "high",
      "metrics": { "avgRating": 4.41, "netStarScore": 74, "negativeReplyRate": 0.82, "invitedShare": 0.12 },
      "percentiles": { "avgRating": 100, "netStarScore": 100, "negativeReplyRate": 100, "invitedShare": 100 },
      "overallPercentile": 91.7
    }
  ]
}

(Trimmed to one company and four metrics. Benchmarked metrics: avgRating, netStarScore, ratingDelta90d, negativeReplyRate, negativeMedianReplyHours, invitedShare, duplicateTextShare, redFlagCount.)

Field reference

Dataset views in the Apify console

• All rows: mixed business, review, search_result, and insight rows
• Reviews: review columns including sentiment
• Businesses & search hits: profiles and discovery results
• Insights: one line per company with rating, trend, reply rate, integrity and signals
• Insight details: themes with evidence, distinctive terms, momentum series, audience
• Benchmark: peer ranking when 2+ companies are analyzed

---

Run it with the API, Python, or JavaScript

cURL

curl "https://api.apify.com/v2/acts/devil_port369-owner~trustpilot-reviews/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{"mode": "companyReviews", "companyUrls": ["apify.com"], "maxReviews": 50}'

Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")
run = client.actor("devil_port369-owner/trustpilot-reviews").call(run_input={
    "mode": "companyReviews",
    "companyUrls": ["apify.com"],
    "maxReviews": 0,
    "sort": "recency",
    "analyzeReviews": True,
})

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    if item["type"] == "insight":
        print(item["executiveSummary"])
        for s in item["signals"]["red"]:
            print(s["severity"], s["code"], s["detail"])

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_API_TOKEN' });
const run = await client.actor('devil_port369-owner/trustpilot-reviews').call({
  mode: 'companyReviews',
  companyUrls: ['apify.com'],
  maxReviews: 0,
  sort: 'recency',
  analyzeReviews: true,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
const insight = items.find((i) => i.type === 'insight');
const reviews = items.filter((i) => i.type === 'review');

Schedule recurring runs under Schedules and connect results to Google Sheets, Slack, Make, n8n, Zapier, or webhooks for continuous reputation tracking.

---

Use with AI agents and MCP

This Actor is built to be called by LLM agents: flat typed output, a small set of enumerated inputs, predictable defaults, and an insight row that already carries the caveats an agent needs to avoid over-claiming.

Connect through the Apify MCP server

Add this Actor to your MCP client (Claude, Cursor, VS Code, and others) through the Apify MCP server:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=devil_port369-owner/trustpilot-reviews",
      "headers": { "Authorization": "Bearer YOUR_API_TOKEN" }
    }
  }
}

Agent decision guide

User intent	Set	Required field
"What do customers say about company X?"	mode: companyReviews	companyUrls
"What's the TrustScore of company X?"	companyReviews, maxReviews: 1	companyUrls
"Is company X's rating trustworthy?" / "Reputation risk?"	companyReviews, analyzeReviews: true, maxReviews: 0, sort: recency	companyUrls
"Compare A, B, C"	as above with several companyUrls	companyUrls (read the benchmark row)
"Find companies that do Y in country Z"	mode: search	searchQueries, country
"List companies in category C"	mode: category	categories
"Find companies and analyze their reviews"	search, scrapeReviews: true, analyzeReviews: true	searchQueries
"Only complaints"	add stars: ["1","2"] (analysis is then disabled for trends)	–
"Only recent"	add date: "last30days"	–

Agent rules of thumb

1. Always set mode first, then the matching array. Each array is a list of strings, even for a single value.
2. Start small, then expand: maxReviews of 20–50 for exploration; use 0 or 500+ for any trend or risk claim.
3. Branch on item.type: business for profile facts, review for review text, insight for analysis, benchmark for rankings, search_result for lead lists.
4. Read sample.confidence and sample.warnings before quoting any insight number. Do not state a trend when momentum.ratingTrend.verdict is insufficient_data.
5. Do not combine analyzeReviews with review filters if you want company-level conclusions. Filtered runs are marked sample.filtered: true and trend verdicts are disabled.
6. Treat integrity flags as leads to investigate, not accusations. Say "flagged for" rather than "fake".
7. Cite evidence: theme evidence entries carry reviewId, date, and a snippet. Every row has profileUrl and scrapedAt for attribution and freshness.
8. stars takes strings (["1","2"]), not integers.
9. Empty email/phone is normal. It means Trustpilot does not publish them for that company.

Minimal agent payloads

{ "mode": "companyReviews", "companyUrls": ["apify.com"], "maxReviews": 50, "includeBusinessProfile": true }

{ "mode": "companyReviews", "companyUrls": ["apify.com"], "maxReviews": 0, "sort": "recency", "analyzeReviews": true }

{ "mode": "search", "searchQueries": ["software company Czech Republic"], "country": "CZ", "maxResults": 25, "scrapeReviews": false }

---

Frequently asked questions

Is scraping Trustpilot legal?

This Actor collects publicly visible data. Legality depends on your use case, jurisdiction, and Trustpilot's terms. Review the Compliance section, avoid using personal data beyond what the law allows, and consult a legal professional for your situation.

Do I need a Trustpilot account or API key?

No. The Actor reads public profile, search, and category pages.

Is this a Trustpilot API alternative?

Yes. It gives you structured review and company data without a Trustpilot business account or API approval process.

How does the Trustpilot sentiment analysis work?

Each review gets a score that blends its star rating with the sentiment of its English text (scored with VADER). The Actor then summarizes the whole sample into an insight row with ratings, trends, themes, reply behaviour, and integrity signals. See How does Trustpilot review sentiment analysis work?.

Does the analysis use an LLM or send my data to an AI service?

No. The analysis runs inside the Actor using rule-based sentiment and statistics. No review text is sent to an external AI service, and no extra API keys are needed.

Which languages does sentiment analysis support?

Text sentiment is English-only. Reviews in other languages are scored from the star rating alone (textSentimentScore is null), and sample.englishTextShare shows how much of the sample had text scored.

What does insufficient_data mean?

The sample cannot support a reliable trend, for example it does not reach back 180 days, has fewer than 20 reviews in a 90-day window, or was filtered. The reason field says which. Raise maxReviews (use 0 for large companies) and remove review filters to get a verdict.

Why were trend verdicts disabled on my run?

Either you used review filters (stars, topics, verifiedOnly, withReplies, reviewSearch), sorted by something other than recency, or the sample was too short. The insight row says which in sample.warnings.

What is the "net star score"?

The share of 4–5 star reviews minus the share of 1–2 star reviews, from −100 to +100. It is more sensitive than the average because it shows how many customers are clearly unhappy.

What is the difference between invited and organic reviews?

Invited reviews are requested by the company (for example via an email link, shown as source: BasicLink and verificationLevel: invited). Organic reviews are written without an invitation. Invited reviews tend to be more positive, so a large gap between them is worth investigating.

Does a red flag mean the reviews are fake?

No. Integrity signals are heuristics, such as an unusual one-day spike or many first-time reviewers. They show where a human should look, and they can have innocent causes like a marketing campaign.

How many reviews do I need for a good analysis?

At least a few hundred for trends, and the more the better. For small companies, the whole profile may be only 100 reviews, and sample.coverage shows how much of the profile you captured. Confidence is low below 100 reviews, medium from 100, and high from 400.

How many reviews can I get per company?

Up to what Trustpilot exposes publicly. Set maxReviews to 0 for the maximum, or use stars, date, and topics filters to cover different slices of a large review set (note that filters disable company-level analysis).

Can I get only negative reviews?

Yes. Use "stars": ["1", "2"]. Combine with date, verifiedOnly, or topics to narrow further.

Can I scrape reviews in other languages?

Yes. Set languages to an ISO code such as es, de, or fr. The default is en.

Why are email and phone empty on some rows?

They are returned only when Trustpilot publishes them for that company. Many businesses hide them.

Why did my search return companies from other countries?

Search follows Trustpilot's index and your country value. Use a more specific query and set country to the market you care about.

What is identifyingName?

The company slug in https://www.trustpilot.com/review/{identifyingName}, usually the domain (apify.com, www.adorama.com).

What is the difference between business and search_result?

business is a full profile from a company's review page. search_result is a lighter card from a search or category listing.

Does scrapeReviews work in company mode?

No. Company mode always scrapes reviews for the URLs you give. scrapeReviews only extends search and category runs.

Can I run it on a schedule and send results to my tools?

Yes. Use Apify Schedules, webhooks, and integrations (Google Sheets, Slack, Make, n8n, Zapier) or pull data through the API.

I found a bug or need extra fields. What should I do?

Open an issue from the Actor's Issues tab. Custom fields, modifications, and private versions are available on request via the DataFusionX contact page.

---

Compliance

• This Actor collects publicly available data. You are responsible for using it in line with Trustpilot's terms and applicable laws such as GDPR and CCPA.
• Review rows include public reviewer names and countries, which can be personal data. Minimize, secure, and retain them only as needed, and honor deletion requests. Insight rows contain no reviewer names; evidence quotes carry only a review ID.
• Insight and benchmark rows are statistical summaries of a review sample for research purposes. They are not investment, legal, or financial advice, and integrity signals are not findings of fraud.
• Not affiliated with, endorsed by, or sponsored by Trustpilot A/S. "Trustpilot" and "TrustScore" are trademarks of their owner.
• This section is general information, not legal advice.

---

More Actors by DataFusionX

Looking for more production-grade scrapers and data pipelines? Browse the full portfolio: apify.com/devil_port369-owner

Need a custom Actor, maintenance, or integrations with Make, n8n, Zapier, CRMs, or Google Sheets? Submit a project request.

---

Keywords: Trustpilot scraper, Trustpilot reviews scraper, Trustpilot sentiment analysis, Trustpilot review analytics, scrape Trustpilot reviews, Trustpilot API alternative, Trustpilot company data, Trustpilot category scraper, Trustpilot search scraper, TrustScore extractor, review sentiment analysis, review authenticity check, reputation monitoring, competitor benchmarking, investor due diligence, voice of customer, customer sentiment data, Trustpilot MCP, AI agent review data