Airbnb / STR Comps + Review Tracker

Public-page Apify Actor for snapshotting Airbnb-like listing data, market/search listing links, visible review snippets, and simple deltas versus a prior snapshot.

The repository is fixture-backed for deterministic offline testing. Live runs only target public HTML pages and do not use login flows, private APIs, or bypass behavior.

Features

• Accepts direct listingUrls and optional marketUrls
• Extracts visible fields only: title, location, rating, review count, visible prices, visible fees, amenities, host name, listing URL, and optional public review snippets
• Discovers listing links from market/search pages up to maxListings
• Emits warning rows for login walls, blocked pages, and fetch failures with evidence
• Computes priceDelta, ratingDelta, reviewCountDelta, isNew, and isMissing against previousSnapshot
• Includes offline fixture smoke coverage

Input

See .actor/input_schema.json or examples/sample-input.json.

{
  "listingUrls": ["https://www.airbnb.com/rooms/1001"],
  "marketUrls": ["https://www.airbnb.com/s/denver--co/homes"],
  "maxListings": 25,
  "previousSnapshot": [],
  "includeReviews": false,
  "fixtureMode": false
}

Output

Dataset rows include:

• listingUrl, sourceUrl, hostUrl
• title, location, hostName
• rating, reviewCount
• pricePerNight, cleaningFee, serviceFee, totalPrice
• amenities, reviews
• status, warningType, warningMessage, evidence
• priceDelta, ratingDelta, reviewCountDelta
• isNew, isMissing, previousStatus
• snapshotTimestamp

Example output is in examples/sample-output.json.

Local Usage

1. Install dependencies:

$npm install

2. Run offline fixture mode:

$npm start -- --input=examples/sample-input.json

Or with Apify locally:

$apify run -i examples/sample-input.json

Tests

npm test
npm run smoke

Both commands run offline and do not depend on live Airbnb pages.

Publish

$apify push

Limitations

• The actor only uses public HTML and visible page content; it does not authenticate.
• Real-world page layouts vary. The extractor is intentionally narrow and may need selector updates if public markup changes.
• Market discovery is limited to links visible in supplied HTML.
• Warning rows indicate blocked/login-gated pages rather than pretending extraction succeeded.