🏢 Glassdoor Scraper — one Actor, three very different jobs

Glassdoor hides jobs, employee reviews, salary percentiles, interview questions and company profiles behind a Cloudflare wall and an account gate. This Actor reads the page data directly — no login, no browser, no cookies — and returns flat rows.

But what you pull, and which fields matter, depends entirely on who you are. So here it is by person.

• 🎯 The job seeker — reviews, interview questions, live listings
• 🧑‍💼 The recruiter & employer-brand team — sub-ratings, sentiment, company profile
• 📊 The comp & market analyst — salary percentiles and pay ranges

Every item comes tagged with a dataType (job, review, salary, interview, or company), so mixed runs stay easy to split.

SOURCE                      WHAT YOU TURN ON               dataType
┌────────────────┐  search   ┌──────────────────────┐
│ 🔑 keywords    │──────────▶│ job listings         │──────────▶ job
└────────────────┘           └──────────────────────┘
┌────────────────┐           ┌──────────────────────┐──────────▶ review
│ 🔗 startUrls   │  per       │ scrapeReviews  (on)  │
│ /Reviews/      │  company   │ scrapeSalaries       │──────────▶ salary
│  ...-E<id>.htm │──────────▶ │ scrapeInterviews     │──────────▶ interview
└────────────────┘           │ scrapeCompanyProfile │──────────▶ company
                             └──────────────────────┘
      ☁️ Cloudflare wall → residential-US proxy · blocked sources → errors

---

🎯 The job seeker

You're deciding whether to apply, and preparing if you do. You care about three things: what employees really say, what the interview throws at you, and which roles are open.

📝 Reviews give you the unfiltered read — summary, pros, cons, advice, the overall rating and all six sub-ratings (ratingWorkLifeBalance, ratingCultureAndValues, ratingCareerOpportunities, ratingCompensationAndBenefits, ratingSeniorLeadership, ratingDiversityAndInclusion), plus recommendToFriend, ceoApproval, businessOutlook, and any employerReply.

🎤 Interviews are the prep goldmine — difficulty, experience, outcome, process durationDays, and the actual questions asked (questionCount tells you how many).

💼 Jobs show what's live — title, location, isRemote, easyApply, ageInDays, and a salary range when Glassdoor has one.

Point it at a company and grab reviews + interviews in one go:

{
  "startUrls": ["https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm"],
  "scrapeReviews": true,
  "scrapeInterviews": true,
  "maxReviewsPerCompany": 100,
  "reviewSort": "recent"
}

Or search for openings by title and location:

{
  "keywords": ["software engineer"],
  "location": "New York, NY",
  "maxJobsPerSearch": 30
}

Tip: set reviewSort to lowest to read the worst reviews first — the fastest way to spot dealbreakers.

---

🧑‍💼 The recruiter & employer-brand team

You're benchmarking reputation — yours against competitors — and you need structured culture signals, not screenshots.

🏢 Company profiles are your scorecard. One row per company: overallRating, reviewCount, recommendToFriendPercent, ceoApprovalPercent, businessOutlookPercent, all six sub-ratings (cultureAndValuesRating, workLifeBalanceRating, careerOpportunitiesRating, compensationAndBenefitsRating, seniorLeadershipRating, diversityAndInclusionRating), plus industry, revenue, size, type, stock, headquarters, founded, website and mission.

📊 Reviews at scale feed theme analysis. Turn on sentiment and each review with text also gets sentimentScore, sentimentLabel and wordCount — so you can quantify pros/cons across hundreds of reviews, not eyeball them.

Pull the profile and sentiment-scored reviews for a set of competitors:

{
  "startUrls": ["https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm"],
  "scrapeReviews": true,
  "scrapeCompanyProfile": true,
  "sentiment": true,
  "maxReviewsPerCompany": 100,
  "reviewSort": "relevance"
}

The Companies tab lays every profile side by side; the built-in HTML report (htmlReport, on by default) renders the sub-ratings, sentiment and pros/cons as an employer-intelligence dashboard saved to the key-value store under report.

---

📊 The comp & market analyst

You size offers and map pay bands. You want percentiles, not anecdotes.

💰 Salaries are reported per job title: baseP10, baseMedian, baseP90, totalMedian, salaryCount (how many reports back the number), currency, mostRecent. That's a distribution you can drop straight into a model.

💼 Jobs add live-market pay: salaryMin / salaryMedian / salaryMax, salaryCurrency, salaryPeriod and salarySource on every listing, so you can compare advertised ranges against reported medians.

Collect salary percentiles by title for a company:

{
  "startUrls": ["https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm"],
  "scrapeSalaries": true,
  "maxSalariesPerCompany": 50,
  "scrapeReviews": false
}

Or run everything at once for a full employer picture:

{
  "startUrls": ["https://www.glassdoor.com/Reviews/Amazon-Reviews-E6036.htm"],
  "scrapeReviews": true,
  "scrapeSalaries": true,
  "scrapeInterviews": true,
  "scrapeCompanyProfile": true,
  "maxItems": 50
}

Salaries, interviews and company profiles are add-ons, off by default — flip them on per company URL. The Salaries tab is pre-built with P10 / median / P90 columns.

---

🌐 The one setup detail everyone needs

Glassdoor is behind Cloudflare, which blocks datacenter IPs. The default proxyConfiguration is preset to residential US and tuned for reliable access — leave it on. Sources that stay blocked or throttled are written to a separate errors dataset with a reason, so your clean data isn't diluted.

Provide at least one source: companies (plain company names), keywords (job search) or startUrls (company URLs like /Reviews/Company-Reviews-E<id>.htm).

A company name is now enough. Reviews, salaries, interviews and the profile are all keyed off Glassdoor's internal employer id, which previously only a pasted URL carried — type "Stripe" and you received job ads and nothing else. Names are matched against Glassdoor's own employer search, so "Stripe" resolves and a job phrase like "software engineer" resolves to nothing and is reported rather than scraped as if it were a company.

---

📥 Full input reference

Field	Default	What it does
companies	["Stripe"]	Company names — the source for reviews, salaries, interviews and profiles
keywords	["software engineer"]	Job titles / keywords to search
location	—	Optional job-search location
startUrls	—	Glassdoor company reviews URLs
scrapeReviews	true	Collect employee reviews per company URL
maxReviewsPerCompany	100	Cap on reviews per company
reviewSort	relevance	relevance · recent · lowest
maxJobsPerSearch	30	Cap on jobs per keyword search
scrapeSalaries	false	Salary percentiles by job title
maxSalariesPerCompany	50	Cap on salary rows per company
scrapeInterviews	false	Interview experiences + questions
maxInterviewsPerCompany	50	Cap on interview rows per company
scrapeCompanyProfile	false	Add a company profile row
minReviewRating · maxReviewRating	—	Keep reviews inside a star band (1–5)
currentEmployeesOnly	false	Keep reviews from current employees only
minReviewWords	—	Drop one-word reviews below this length
reviewsWithinDays	—	Keep reviews published within N days
minJobSalary	—	Keep jobs paying at least this
jobLocationType	ANY	REMOTE · ONSITE · HYBRID
jobsWithinDays	—	Keep jobs posted within N days
minSalarySamples	—	Drop salary stats built on a tiny sample
interviewDifficulty	ANY	EASY · MEDIUM · HARD
interviewsWithinDays	—	Keep interviews reported within N days
excludeKeywords	—	Drop any record whose text contains a phrase
sentiment	false	Sentiment score + label on reviews
htmlReport	true	Build the shareable HTML report
dedupe	true	Save each job/review once per run
maxItems	200	Global cap across all items (0 = unlimited)
proxyConfiguration	Residential US	Residential required (Cloudflare)
maxConcurrency · maxRetries · requestTimeoutSecs	5 · 5 · 30	Request tuning

Filters keep the noise out. Every quality filter runs on a finished record before it is saved, so anything it drops is never charged. A filter only tests the record type it belongs to — a rating filter never touches a salary row — and a record missing the value a filter checks is kept, so filtering never looks like data loss. Each run reports filteredOut, filterBreakdown per setting, and a plain-language filterHint naming the costliest one.

---

🆕 What each row now carries

Reviews hand back all six sub-ratings as their own columns — work-life balance, culture, career, pay and benefits, leadership, diversity — next to CEO approval, business outlook and recommend-to-friend. An unrated dimension stays null instead of vanishing, so a CSV export keeps stable columns. Each row also says whether it came from a former employee, which tenure band they sat in, whether the employer replied and when.

Company profiles are scored against their own industry, not in a vacuum: industryAverageRating, ratingVsIndustry and beatsIndustryAverage turn "3.6 stars" into an argument. They also carry the content counts — reviews, salaries, interviews, photos, benefits, open jobs — plus office locations, sector and whether the profile is employer-managed.

Salaries add the total-pay percentiles beside base pay, the width of the band (basePaySpreadPercent) and how much of the median package is not base salary (additionalPayMedian, additionalPaySharePercent).

Interviews keep the real questions asked, and add how the candidate applied, how many days the process ran, and a plain gotOffer flag derived from the outcome.

🗂️ Datasets & tabs

The run writes to one dataset with ready-made views — Overview (quick-scan mix), Jobs, Reviews, Salaries, Interviews, Companies — plus a separate errors dataset for skipped sources.

---

🙏 If this Actor saved you a login wall or two, a short review on the Apify Store goes a long way — it helps other job seekers, recruiters and analysts find it, and points me at what to improve next.

🏷️ Tags: glassdoor · glassdoor scraper · jobs scraper · company reviews · salary data · interview questions · employer branding · hr analytics · recruitment