Trustpilot Reviews Scraper: Company Reviews, Replies, Alerts

Get public Trustpilot reviews for any company, as clean JSON, CSV or Excel. Filter by stars, date and language, read past the first 200 reviews, and get only new reviews on each scheduled run.

Capterra (experimental): Capterra blocks most automated access, including from Apify's servers in our tests on 21 October 2026. Capterra targets may return a free blocked status instead of reviews. You are never charged for a blocked target.

You give it a list of companies or products. You get back:

• One row per review: rating, title, text, pros and cons, date, language, reviewer display name and country, verified flag, company reply and helpful votes.
• One free profile row per company: overall rating, review count, star distribution, categories, country and how often the company replies.
• One free status row per target, so you can see what worked and what did not.

No login. No captcha solving.

Who uses it

• Marketing teams that track what customers say about their brand and about competitors.
• Product teams that read Capterra pros and cons to plan features.
• Reputation agencies that need new reviews for many clients every day, with the company reply next to each review.
• AI and sentiment pipelines that need clean review text, a star label and a language code, in one shape for every source.

What you can put in

One target per line. Mix them freely.

You type	What it means
https://www.trustpilot.com/review/careem.com	Trustpilot company page
careem.com	Same thing, shorter
Talabat	A company name. The actor searches Trustpilot and takes the best match. The status row says which company it matched.
https://www.capterra.com/p/135003/Slack/reviews/	Capterra product page
https://www.g2.com/products/slack/reviews	Accepted, returns a free not_supported status row. See Limits.

Other options:

• Max reviews per target. You pay only for reviews delivered.
• Star ratings. For example only 1 and 2 stars.
• Reviews published on or after. A date like 2026-01-01, or 30 days, or 6 months.
• Review language. all, or a code like en, ar, de.
• Order. Newest first, or Trustpilot's "most relevant".
• Only new reviews since the last run. Turn this on for scheduled alerts. Give each watch list its own state name.
• Company profile row. On by default, and free.

Output

Review row (Trustpilot, real run 2026-10-21)

{
  "type": "review",
  "source": "trustpilot",
  "targetName": "Careem",
  "targetUrl": "https://www.trustpilot.com/review/careem.com",
  "targetRating": 1.1,
  "targetReviewCount": 1066,
  "reviewId": "69946cff1d678738ce2084f7",
  "reviewUrl": "https://www.trustpilot.com/reviews/69946cff1d678738ce2084f7",
  "rating": 5,
  "title": "Amazing  very very good",
  "text": "Amazing  very very good ",
  "pros": null,
  "cons": null,
  "date": "2026-02-17T15:28:31.000Z",
  "experienceDate": "2026-02-05T00:00:00.000Z",
  "updatedDate": null,
  "language": "en",
  "languageSource": "site",
  "reviewLength": 46,
  "reviewerName": "Sura",
  "reviewerCountry": "AE",
  "verified": false,
  "companyReplyText": "Wow! Thank you so much for the amazing feedback. We’re thrilled you had such a good experience!",
  "companyReplyDate": "2026-02-20T12:05:38.000Z",
  "helpfulVotes": 0,
  "subRatings": null,
  "input": "careem.com",
  "scrapedAt": "2026-10-21T14:32:17.326Z"
}

Arabic reviews come through as written, with "language": "ar":

{"targetName": "Careem", "rating": 3, "title": "السواق محترم والسيارة نظيفة", "text": "السواق محترم والسيارة نظيفة، بس اتأخر شوي بسبب الزحمة.", "language": "ar", "date": "2025-11-06T23:35:33.000Z", "reviewerCountry": "IQ"}

Review row (Capterra, parsed from a real page saved 2026-10-21)

Capterra rows add pros, cons and subRatings. Text is shortened here.

{
  "type": "review",
  "source": "capterra",
  "targetName": "Slack",
  "targetUrl": "https://www.capterra.com/p/135003/Slack/reviews/",
  "targetRating": 4.7,
  "targetReviewCount": 24230,
  "reviewId": "7209002",
  "rating": 5,
  "title": "Slack: Essential for remote work productivity...and employee engagement ",
  "text": "Slack has been part of the corporate toolkits of the companies I've worked for since 2017 ...",
  "pros": "Since 2020, I've worked remotely as a business developer across two digital services companies. ...",
  "cons": "Honestly, not much. The only time I get frustrated with it is when I need to invite partners ...",
  "date": "2026-09-03T00:00:00.000Z",
  "language": "en",
  "languageSource": "detected",
  "reviewerName": "Alicia R.",
  "reviewerCountry": null,
  "verified": true,
  "companyReplyText": null,
  "subRatings": { "easeOfUse": 5, "customerSupport": 5, "functionality": 5, "valueForMoney": 5, "likelihoodToRecommend": 10 }
}

Field notes

• date is the publish date in ISO format. Capterra gives only a day, so its time is midnight UTC.
• language: Trustpilot sets its own language tag (languageSource: "site"). For Capterra the actor checks the text with a simple built-in rule: alphabet first, then common words (languageSource: "detected"). It returns null when it is not sure.
• reviewLength counts the characters of title, text, pros and cons together.
• verified: Trustpilot's "verified" label, or Capterra's "validated reviewer" check.
• helpfulVotes: Trustpilot "useful" votes. Capterra does not show a count, so it is null.
• subRatings: Capterra only. A sub-rating the reviewer skipped is null.
• Personal data: the actor keeps only the reviewer's display name and country. It drops profile photos, user IDs, job titles and employer details.

Company profile row (free)

{
  "type": "company",
  "source": "trustpilot",
  "name": "Careem",
  "domain": "careem.com",
  "url": "https://www.trustpilot.com/review/careem.com",
  "website": "https://careem.com",
  "trustScore": 1.1,
  "reviewCount": 1066,
  "reviewCountLast12Months": 209,
  "ratingDistribution": { "1": 992, "2": 29, "3": 10, "4": 8, "5": 27 },
  "categories": ["Courier and Delivery Service", "Taxi Service", "Payment Service"],
  "country": "AE",
  "isClaimed": true,
  "averageDaysToReply": 6.31,
  "replyPercentage": 73.89162561576354,
  "charged": false
}

Status row per target (free)

{"type": "status", "source": "trustpilot", "input": "careem.com", "targetName": "Careem", "status": "ok", "reviews": 250, "pages": 15, "newestReviewDate": "2026-10-20T12:51:45.000Z", "oldestReviewDate": "2017-08-05T07:57:32.000Z", "seconds": 18.9, "charged": false}

Status values:

• ok: reviews delivered.
• no_reviews: the page exists but nothing matched your filters.
• no_new_reviews: "only new" mode found nothing since the last run.
• not_found: no company page for this domain or name.
• blocked: the site refused every retry on every proxy tier.
• blocked_captcha: the site asked for a captcha. The actor does not solve captchas.
• not_supported: G2 (see Limits).
• bad_input: the line is not a URL, domain or name the actor understands.
• skipped_cost_limit or cut_by_cost_limit: your maximum cost for the run was reached.

The RUN_SUMMARY record in the key-value store has every status row, reviews per source, requests, bytes per 1,000 reviews and the proxy tier used.

How it works

• Trustpilot. Each company page holds its reviews as JSON inside the page. The actor reads that JSON. Some networks get Trustpilot's automatic "Verifying your connection" page first. In that case the actor opens that one page in a headless browser, the site sets its normal cookie, and the actor goes on with plain HTTP. The browser loads only the check page and closes after a few seconds.
• More than 200 Trustpilot reviews. Logged-out visitors can open 10 pages (200 reviews) for each filter. When you ask for more, the actor reads each star rating separately, then each language, and removes duplicates. In a test it read 450 Amazon reviews in 22 seconds this way.
• Capterra. The reviews sit inside the page data. The actor reads 25 reviews per page and follows ?page=2, ?page=3 and so on.
• Proxies. The actor starts without a proxy. If a site keeps refusing, it moves to Apify datacenter proxy, then residential. You can turn residential off, or set your own proxy.

Pricing

$1.00 per 1,000 reviews ($0.001 per review). Each run also has a start fee of $0.00005 (per GB of memory). Company profile rows and status rows are free. Blocked and unsupported targets cost nothing. Set a maximum cost on the run and the actor stops cleanly when it is reached.

Measured on 2026-10-21 for Trustpilot (local runs, 1,450 reviews): about 7.7 MB of compressed traffic and 59 requests per 1,000 reviews, about 25 seconds of run time per 1,000 reviews at 256 MB.

Limits

• G2 is not supported. G2 shows a captcha to every logged-out visitor, including a normal desktop browser. This actor does not solve captchas. G2 URLs return a free not_supported row.
• Capterra is protected by a Cloudflare check. From our test network every automated request got the check page, so Capterra live runs were not confirmed. The parser is tested on a real saved page. If Capterra blocks every proxy tier, you get a free blocked row and pay nothing.
• Capterra order. Capterra pages come in its "most helpful" order. Date, star and language filters are applied to the pages the actor reads, so a narrow filter can need many pages.
• Capterra has no reviewer country and no helpful count.
• Trustpilot depth. Even with splitting, each star and language combination gives at most 200 reviews. Companies with very many reviews in one language and one star rating cannot be read in full while logged out. The status row says so.
• Trustpilot browser check. The actor passes the automatic check page only. If Trustpilot shows a captcha instead, the target ends as blocked_captcha.
• Name search is Trustpilot only. For Capterra, paste the product URL.

Tips

• For daily alerts, schedule the actor with "Only new reviews" on, newest first, and a small max. The first run sets the baseline.
• To compare competitors, put all of them in one run. Filter the dataset by targetName.
• To feed an AI pipeline, keep text, pros, cons, rating and language. They are filled the same way for both sites.