Glassdoor Reviews Scraper & API

Get every employee review of any company on Glassdoor, in bulk. Paste company links or names. Each review comes back with the overall rating and the six sub-ratings (work/life balance, culture and values, diversity and inclusion, career opportunities, compensation and benefits, senior management), recommend to a friend, CEO approval, business outlook, headline, pros, cons, advice to management, job title, location, employment status, current or former employee, time at the company, helpful votes, language, date, and the company's responses. Filter by date, rating, language, job title, location or current employees. Export as JSON, CSV or Excel, or read it through the Apify API.

No reviewer name, profile or ID. We deliver what was said about the company, not who said it. Names and signatures in employer replies are removed.

Quick start

1. Paste one or more company links into Glassdoor links (https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm, from any Glassdoor country site), or names into Company names (Sixt).
2. Set Max reviews per company (empty reads them all) and Max results in total.
3. Click Start and read the reviews in the run's Output tab.

Like any Apify Actor, it can also be run and read through the Apify API, scheduled, and connected to webhooks, Make, Zapier, n8n or Apify's MCP server for AI agents.

What it is used for

• Employer reputation: what staff say about your company, or only the reviews since last week, every day.
• Competitor and talent research: pay, culture and management ratings of the companies you hire against.
• HR analytics and market studies: one chain across cities, job titles and years, with the six sub-ratings.
• Sentiment analysis and AI training: one clean record per review, pros and cons apart, the company's response next to it, a stable ID to deduplicate between runs.

What you get

Review ID (unique on Glassdoor: use it to deduplicate between runs)	yes
Overall rating (1 to 5), headline, pros, cons	yes
Advice to management	when given
Six sub-ratings (1 to 5)	when the reviewer rated them
Recommend to a friend, CEO approval, business outlook	when answered
Job title, location	when given
Employment status (full-time, part-time, contract, intern…), current or former employee	yes
Time at the company (less than 1 year, more than 3 years…)	when given
Helpful and not helpful votes	yes
Language of the review (en, fr, de, nl, pt, es, it)	yes
Date of the review	yes
The company's responses: text and date	when there is one
Link of the review and of the company, company name and ID	yes
Reviewer name, profile, ID	no, never
Names and signatures in the company's responses	removed: a person's sign-off (name, with the job title and team lines under it), a name in the greeting, a person named in the text ([name]). A sign-off with only the team or company name stays. An email address is kept only when it is a service address such as hr@, jobs@, careers@

Name removal is good, not perfect. It is done by a name recognition model plus layout rules, tested on about 2,700 written cases and 467 real replies; we found no leaked name in the standard layouts, and a few leaks or damaged words in the hardest sets, so we do not promise zero. Known weak spots: a first name that is also a common word or the company's own name, a name written in a script the model cannot read (Cyrillic, Greek, Korean, Japanese, Chinese, Arabic, Hebrew, Thai) in the middle of a sentence rather than in a greeting or sign-off, a lone initial, and unusual sign-off layouts. Teams and company names are kept on purpose.

A field Glassdoor does not give for a review is null, never missing. A sub-rating the reviewer skipped is null (never 0).

Inputs

• Glassdoor links: any page of the company (reviews, overview, salaries, jobs) on any Glassdoor site. A link that already filters a city or a state keeps that filter.
• Company names: as written on Glassdoor. A name that matches exactly one company is used; a name that matches several gets a free error record that lists them, so you can pick the link.
• Max reviews per company (empty: all) and Max results in total (1,000 by default: raise it when you want more; Amazon alone has over 250,000 reviews). Both are counted exactly, also when a company is given both by link and by name.
• Order: newest first (default) or oldest first.
• Since and Until: a date such as 2026-01-31, or a duration such as 30 days. The reading starts and stops at your dates, without reading the years in between. Glassdoor keeps one list per language, so a date far in the past is looked up in each of them: on a very large company (Amazon), reading this far back in every language can cost more than one run can cover. The run then says so; pick one or two review languages.
• Minimum rating and Maximum rating: 1 to 5 stars.
• Languages: en, fr, de, nl, pt, es, it, one per line. Empty: every review.
• Job title: as Glassdoor writes it (Software Engineer, Cook). Glassdoor includes closely related titles (Software Developer with Software Engineer). A title Glassdoor does not know gets a free error record with the closest titles it lists.
• Location: a city, state or country as Glassdoor writes it (Seattle, WA, Texas, Germany). One that matches several places gets a free error record that lists them.
• Current employees only, and Review contains (up to 50 words, one per line, matched as whole words in the headline, pros, cons and advice).

Example, through the API:

{
  "startUrls": ["https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm"],
  "companyNames": ["Sixt"],
  "since": "90 days",
  "maxRating": 2,
  "location": "Seattle, WA",
  "maxReviewsPerCompany": 500
}

Output

One record per review in the dataset. A shortened review:

{
  "type": "review",
  "status": "ok",
  "reviewId": "105785119",
  "companyId": "10875",
  "companyName": "Sixt",
  "rating": 2,
  "title": "Bonnes primes, mais management limité et heures supp négligées",
  "pros": "Grosses primes si bon commercialement",
  "cons": "Cela dépend sur quel secteur mais management très limite et heures sup…",
  "advice": null,
  "recommend": "negative",
  "ceoApproval": null,
  "businessOutlook": "positive",
  "workLifeBalanceRating": 3,
  "cultureAndValuesRating": 2,
  "diversityAndInclusionRating": 5,
  "careerOpportunitiesRating": 3,
  "compensationAndBenefitsRating": 2,
  "seniorManagementRating": 2,
  "jobTitle": "RSA",
  "location": "Paris",
  "employmentStatus": "full-time",
  "currentEmployee": false,
  "tenure": "more than 1 year",
  "helpfulVotes": 0,
  "notHelpfulVotes": 0,
  "language": "fr",
  "date": "2026-09-29T20:58:03Z",
  "employerResponse": "Chère ancienne collaboratrice / Cher ancien collaborateur,\n\nMerci pour…",
  "employerResponseDate": "2026-10-05T06:51:03Z",
  "employerResponses": [{"text": "Chère ancienne collaboratrice / Cher ancien collaborateur,\n\nMerci pour…", "date": "2026-10-05T06:51:03Z"}],
  "reviewUrl": "https://www.glassdoor.com/Reviews/Employee-Review-Sixt-RVW105785119.htm",
  "companyUrl": "https://www.glassdoor.com/Reviews/Sixt-Reviews-E10875.htm",
  "input": "Sixt",
  "position": 1,
  "scrapedAt": "2026-10-22T05:40:32Z"
}

• status: ok (delivered, charged) or error (free; errorCode and error say why, retryable says if a later run may work).
• recommend is positive or negative; ceoApproval is approve, disapprove or no_opinion; businessOutlook is positive, neutral or negative.
• language is the language of the review. Glassdoor sorts reviews in seven languages; the very few written in another language come with null (3 of Amazon's 254,716 reviews).
• position is the rank of the review in the chosen order for that company.

The run summary (counts per company, name, Glassdoor's overall rating and review count, why the run stopped, notes) is the OUTPUT record of the run's key-value store.

Pricing

Pay per event. Reviews cost less on higher Apify plans:

Apify plan	Review
Free	$4.30 per 1,000
Starter (Bronze)	$4.30 per 1,000
Scale (Silver)	$4.20 per 1,000
Business (Gold)	$4.10 per 1,000
Platinum	$4.10 per 1,000
Diamond	$4.10 per 1,000

On every plan, the run start is $0.005 per GB of memory (1 GB minimum, which is also the default memory). So 1,000 reviews of one company come to $4.305 on the Free plan and $4.105 on Business.

Never charged: reviews that do not match your filters, error records, repeats (a review already delivered in the same run, for example a company given both by link and by name, is skipped). Set Maximum cost per run in the run options and the run stops before going over it, with the results delivered so far.

Limits

• Every review: the Actor reads the whole history of a company, 100 reviews at a time. On Amazon, about 20,000 reviews in a few minutes.
• Several ratings, words: one rating (for example 1 star only) is filtered by Glassdoor itself; several ratings and Review contains are filtered here, so a rare word on a large company means reading many pages. Reviews that do not match are not charged. Set Max reviews per company or Since to bound it.
• Dates are in UTC, converted from the US Pacific time in which Glassdoor writes them.
• Companies with similar names: use the link when a name is ambiguous.
• If a run stops early (your maximum cost, Max results in total, its timeout), every company it did not read in full and gave nothing for gets a free not_read record, so no input is ever left without an answer.
• Anything missing from a result is said: in the run's status, in the note of each company in the OUTPUT record, or in a free error record. If Glassdoor limits access during a run, the run says so and keeps what it delivered; running it again later gives the rest.

FAQ

Do I need a Glassdoor account or API key? No.

Why are reviewer names not included? Glassdoor reviews are anonymous, and we add nothing that could point to a person: no profile, no ID. Names and signatures in employer replies are removed. The review ID and link let you find any review on Glassdoor.

What if the run is restarted by the platform (migration or reboot)? It resumes where it was: results already in the dataset are not delivered or charged again. If it cannot (its saved progress is missing), it stops at once and says so: what is in the dataset is not charged again, and a new run gives the rest.

Is it legal? It collects public reviews shown to any visitor of Glassdoor, without any personal data. Check the rules that apply to your use of the data.

Does the Actor send anything to Quarra? Only when a run fails or runs out of time: the Actor, its build, the error type, the line of our code where it happened, and the memory and time the run was given, so we can fix it fast. Never your input, your results, your account or the run ID.