📅 Airbnb Availability & Rate Scraper

Scrape the forward-looking availability calendar of Airbnb listings and the real per-day fields Airbnb's own calendar returns — check-in / check-out eligibility, the bookable flag, minimum / maximum nights, and the nightly price when it is present. From those real values it derives an honest unavailableRate and, only from real calendar prices, ADR / estimated revenue / RevPAR.

This actor is deliberately honest about what the data is and is not. Read the two notes below before using it.

⚠️ Honesty notes (read these)

• This is availability, not occupancy. Airbnb's calendar tells you whether a night is available. An unavailable night can be either genuinely booked or blocked by the host — the public calendar cannot distinguish the two. We therefore report unavailableRate = unavailable days ÷ counted days, clearly labeled "includes host-blocked days; not a booking/occupancy rate." We never call unavailable nights "booked" and never emit an "occupancy rate."
• Money fields come only from real prices. adr, estimatedRevenue and revpar are computed exclusively from nightly prices that appear in the calendar payload. When a listing exposes no calendar prices, these fields are null — never a regional average, a search-page price, or an interpolated guess. estimatedRevenue / revpar additionally assume every unavailable night is a paid booking, which is an upper bound (it includes host-blocked nights); they are labeled as such.
• Forward-only. PdpAvailabilityCalendar returns future availability. Past dates are clamped to today. There is no historical calendar available from this endpoint.

How it works

1. curl_cffi Chrome TLS impersonation. Airbnb sits behind Akamai / PerimeterX. Plain requests/aiohttp are blocked at the TLS (JA3) layer before reaching the app. This actor uses curl_cffi with impersonate="chrome" so requests replay a real Chrome handshake.
2. Soft-block guard + proxy escalation. An airlock / challenge / empty shell is detected and reported as a block (the run keys on the presence of the expected bootstrap container, not on loose tokens), and the actor retries on a fresh residential proxy IP.
3. Dynamic persisted-query discovery. The GraphQL persisted-query hash is found by scanning the room page and its JS bundles for the hash next to PdpAvailabilityCalendar — no hardcoded bundle name that breaks on every Airbnb deploy.
4. Harvest + derive. Every already-in-payload day field is emitted; derived metrics are computed with the honesty rules above.

Input

Field	Type	Notes
listings	array (required)	Room URLs or bare IDs, one per line.
startDate	date	Absolute (2026-08-01) or relative (7 days, 1 month). Empty = today. Past → clamped to today.
endDate	date	Absolute or relative. Empty = start + 30 days. Up to ~12 months ahead.
billingMode	enum	per_listing (default, 1 charge/listing) or per_day.
includeDays	boolean	Emit per-day child rows (default true).
includeListingDetails	boolean	Best-effort title/type/capacity/beds/baths/coords (default false).
includeHostProfile	boolean	Best-effort superhost/name/response rate (default false).
proxyConfiguration	proxy	Default residential (recommended; datacenter is usually blocked).

{
  "listings": ["https://www.airbnb.com/rooms/35329051"],
  "startDate": "2026-08-01",
  "endDate": "2026-10-31",
  "billingMode": "per_listing",
  "includeDays": true,
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Output

Two row types share the dataset and are distinguished by type / isChild.

Parent — type: "listing" (one per target):

{
  "type": "listing", "isChild": false,
  "roomId": "35329051", "listingUrl": "https://www.airbnb.com/rooms/35329051",
  "windowStart": "2026-08-01", "windowEnd": "2026-10-31",
  "totalDays": 92, "availableDays": 40, "unavailableDays": 52,
  "unavailableRate": 0.5652,
  "adr": 128.5, "priceCoverage": "40/92", "pricedDays": 40,
  "estimatedRevenue": 6682.0, "revpar": 72.63,
  "minPrice": 110.0, "maxPrice": 210.0,
  "currency": "USD", "scrapedAt": "2026-07-03T12:00:00Z"
}

Child — type: "day" (one per calendar day, mirrored to a per-run days-<runId> dataset):

{
  "type": "day", "isChild": true, "parentId": "35329051", "roomId": "35329051",
  "date": "2026-08-01", "dayOfWeek": "Saturday", "isWeekend": true,
  "available": false, "availableForCheckin": false, "availableForCheckout": true,
  "bookable": false, "minNights": 2, "maxNights": 365,
  "price": null, "currency": null, "scrapedAt": "2026-07-03T12:00:00Z"
}

price is a real number or null (never 0). unavailableRate, adr, estimatedRevenue, revpar are null when the underlying data is absent.

Billing

Pay-per-event. per_listing (default) charges one listing_result per listing. per_day charges one day_result per calendar-day row (~30 for a month, ~365 for a year). Day rows are always produced; billingMode only controls what is billed. See .actor/pay_per_event.json.

Keywords

airbnb scraper, airbnb availability, airbnb calendar, airbnb rates, nightly price, adr, revpar, short-term rental data, str analytics, vacation rental availability, airbnb api, booking calendar, occupancy alternative, curl_cffi, akamai bypass.

Compliance

Only publicly visible calendar data is accessed. Ensure your use complies with Airbnb's Terms of Service and applicable law.