Booking Hotel Finder & Scraper: AI best-value shortlist

The best-value Booking.com hotels for your dates, budget and brief — ranked and explained.

⚡ First 10 hotels free · 💵 $0.004 per qualified hotel · 🤖 $0.01 with AI assessment · 🔑 No API key where your Apify plan includes model access — otherwise your own key · ⏱️ free demo in 3 s

Try it in 30 seconds. Click Try it — the form is pre-filled with a free demo on six sample Lisbon hotels. Type your destination and press Start to search Booking.com live (a Store Actor on your Apify plan); your first 10 qualified hotels are free. → What a run costs · What the AI adds · Run it weekly

⚡ At a glance

🎯 What this Actor does

This booking hotel finder runs a Booking scraper on your account, drops junk for free and hands you a shortlist you can send to a client:

• Price per night you can compare — Booking's stay price ÷ nights, next to rating and reviews.
• Verified facts — amenities from Booking's facility list, free cancellation and breakfast from the room options.
• Fine print before it bites — city tax, resort fee, deposit, cash only, late check-in fees, quoted verbatim.
• Price-drop alerts — scheduled runs deliver only new hotels or cheaper rates, even with relative dates.

🏷️ Best value hotels: the value score

valueScore (0–100) weighs guest rating shrunk by review count (35 %), price per night against this search (30 %), Booking's Value for money / Location / Cleanliness (15 %), your amenities (10 %), free cancellation (5 %) and distance to nearPoint (5 %). valueReasons says why; minScore is the bar. Free filters: rating, stars, type, budget, amenities, free cancellation, breakfast, distance, sold-out rooms and duplicates.

📋 A hotel shortlist your client can read

whyThisHotel is one line per hotel, pros / cons come from Booking's review summary, shortlistMemo: true writes a top-5 comparison, and every run writes REPORT — a client-ready page whose link, OUTPUT.reportUrl, you forward.

🔔 Hotel price drop alert and hotel price tracker

Each delivered stay and its price is remembered (hotel + nights + adults when the dates are relative, 30 days). With onlyNewOrPriceDrop a scheduled run delivers only new hotels or drops of minPriceDropPct % or more, with previousPrice, priceChangePct and a Slack-ready line: ↓12 % Hotel X, $143/night, 8.7 Excellent.

📂 Already scraping Booking.com? Rank your Booking scraper data

Point datasetId at a finished Booking scraper run (or paste rows into itemsList): no second scrape, you pay only per qualified hotel, rules only by default.

🚀 How to use it

1. Click Try it — the form is pre-filled with the free demo on sample Lisbon hotels; it works as is.
2. Type your destination, checkIn and checkOut — the run switches to the live search by itself, capped by maxDiscoveryChargeUsd. Or pick a ready search in preset; add a travellerBrief, keep enableAi on.
3. Press Start — the Shortlist view is ranked best value first.
4. Schedule it daily or weekly with your address in notifyEmail — only new hotels and price drops are charged.

📚 Learn more: Step-by-step tutorial — find the best-value Booking.com hotels with this Actor

🆚 Why this instead of a Booking Scraper?

💸 Cheaper than a rate-shopper seat

Lighthouse (formerly OTA Insight) Rate Insight ($250/month) and RateGain Rate Shopper ($200/month) are per property (indicative). A daily watch here — 100 properties, ~5 new or cheaper a day with AI — is ≈ $0.50 source + 5 × ($0.01 + ≈ $0.003 tokens) = $0.57 a day, ≈ $17/month, no seat.

💵 Pricing — what a run really costs

Pay per result: one event per qualified hotel, nothing else. No start fee.

Never charged by us: hotels below minScore or failing a filter, duplicates, your exclude list, unchanged hotels from an earlier run. The source run is billed by the Booking scraper on your account.

How that compares — $0.004 is the median row price of the Booking scrapers and of the Store search next to this card (one Actor per publisher) — the same price per unit, but one qualified hotel replaces 4 raw rows you would read and discard: $0.001 per raw-row equivalent. Junk costs nothing here.

Worked example — 100 properties scraped = $0.50 on your account → 25 qualify → 10 free + 15 × $0.004 = $0.06 on your first run. Next run with AI on: 25 × $0.01 + 25 × ≈ $0.003 tokens = $0.33 here + $0.50 source = $0.83, $0.033 per hotel (AI off: $0.10 + $0.50 = $0.60). With the $0.003 scraper the source is $0.30: $0.63 all in (AI on), $0.40 (AI off).

AI tier, all in: ≈ $0.013 per hotel ($0.01 + ≈ $0.003 tokens at ~1,600 in / 300 out with anthropic/claude-haiku-4.5 through Apify's model access) — 2.5× the basic event; with your own key the tokens go to it.

🤖 What the AI tier adds

The same demo hotel, AI off — facts, value score, Booking's own pros/cons and a keyword brief match:

{"name": "Baixa Garden Suites", "pricePerNight": 149.0, "rating": 8.9, "valueScore": 68, "briefMatchScore": 100,
 "whyThisHotel": "Rated 8.9 Fabulous from 640 reviews; $149/night, cheaper than 50% of this search; Value for money 8.6, Location 9.6, Cleanliness 9.1.",
 "briefMatched": [{"requirement": "pool", "evidence": "Rooftop pool"},
                  {"requirement": "free cancellation", "evidence": "free cancellation on a room option"}],
 "briefUnmet": [], "aiModel": null, "assessedBy": "rules"}

AI on — judged against "Quiet hotel for a couple, walkable to the old town, with a pool and free cancellation":

{"name": "Baixa Garden Suites", "pricePerNight": 149.0, "rating": 8.9, "valueScore": 72, "briefMatchScore": 80,
 "briefMatched": [{"requirement": "pool", "evidence": "shared rooftop pool"},
                  {"requirement": "walkable to the old town", "evidence": "300 metres from Rossio Square"}],
 "briefUnmet": [{"requirement": "Quiet hotel for a couple", "evidence": null}],
 "whyThisHotel": "Fabulous 8.9-rated apartments close to Rossio Square, featuring a shared rooftop pool and free cancellation.",
 "finePrintFlags": [{"type": "late_checkin_fee", "quote": "Guests arriving after 20:00 must contact the property in advance; late check-in carries an extra charge of EUR 30."},
                    {"type": "city_tax", "quote": "This property only accepts cash for the city tax."},
                    {"type": "party_restriction", "quote": "Parties/events are not allowed."},
                    {"type": "payment_restriction", "quote": "This property only accepts cash for the city tax."}],
 "aiModel": "google:gemini-3.5-flash-lite", "assessedBy": "ai"}

Local run with a Google key, 2026-10-16 (Apify's model access uses anthropic/claude-haiku-4.5). The rules counted "pool" and "free cancellation" as a 100 % match; the AI proves "walkable" with a quote, marks "quiet" unproven and flags the cash-only city tax — ≈ 4 min of reading saved per hotel (estimate on 5 sample hotels).

⚙️ Input

Budget, guests, minRating, onlyNewOrPriceDrop, sendAlertsTo and sourceMode are in the Input tab; webhook headers / signing, dedupeStoreName, language and the run limits are API-only fields.

📤 Output

One dataset item per qualified hotel — JSON, CSV or Excel, or a webhook.

Output fields

Upsert key for a CRM or Sheet: dedupeKey — Booking hotel id + check-in + check-out + adults (hotel + nights + adults with relative dates), the key the price memory uses; itemId is its short hash, hotelId the property.

Views: Shortlist, Price drops, Client sheet, Score breakdown; DIGEST (top 10, trend vs last run) and OUTPUT (funnel, aiCoverage, timing) sit next to the dataset.

⭐ Found it useful? A review on the Store helps others find it — it takes a minute on the Actor's page.

🔁 Run it weekly

1. Actions → Schedule (daily for price-drop alerts) with your address in notifyEmail — each run with new hotels or drops e-mails you the digest and the trend vs the last run. Relative dates (30 days) keep working.
2. Keep dedupeAcrossRuns and onlyNewOrPriceDrop on — you are charged only for what is new or cheaper; the memory is kept per search (destination, guests, source), so a reworded brief keeps it.
3. Paste a Slack, Teams, Google Chat, Discord or Telegram link — or your webhook — into sendAlertsTo, and track the picks in the ready tracker sheet (dedupeKey first to upsert on).

Weekly cost is the new hotels only: 100 properties, ~10 new or cheaper a week, AI on ≈ $0.50 source + 10 × $0.013 all in = $0.63.

🎯 Try it for your niche

Pick one in the form's preset select, or paste its input into the JSON tab:

🔌 Integrations, automation and API

• E-mail / chat — notifyEmail and sendAlertsTo (Slack, Teams, Google Chat, Discord, Telegram) deliver the DIGEST after every run with new hotels.
• Webhook (Zapier, Make, n8n) — a webhook URL in sendAlertsTo gets {"event": "hotel.qualified", "hotel": {…}} per hotel; webhookSecret signs each POST (X-Signature: sha256=…). A failed POST is retried once, listed in OUTPUT.webhook.failedKeys and not re-sent next run — the run never fails for it.
• After your voyager/booking-scraper run — in its Integrations tab, on Run succeeded start this Actor with datasetId = the run's default dataset and enableAi: false: your Booking scraper data, ranked by rules.
• Hotel search API for apps and AI agents / MCP — plain typed fields; the free first call is

  "sourceMode": "list"
  with the Booking records you send.

from apify_client import ApifyClient

client = ApifyClient("<YOUR_API_TOKEN>")
run = client.actor("rich_minds/booking-hotel-finder-ai").call(
    run_input={
        "sourceMode": "list",  # free trial; "actor" + "destination" + dates searches Booking.com live
        "itemsList": [{"name": "Hotel Example", "url": "https://www.booking.com/hotel/pt/example.html", "price": 300,
                       "checkInDate": "2026-10-26", "checkOutDate": "2026-10-28", "rating": 8.8, "reviews": 950}],
        "travellerBrief": "Quiet couple trip with a pool",
        "maxDiscoveryChargeUsd": 0.5,  # spend cap once you switch to "sourceMode": "actor"
    },
    timeout_secs=1800,  # a live search with AI — see "How long does a run take?"
    # build="0.1.8",  # pin the build your pipeline was tested with (Changelog below)
)
out = client.key_value_store(run["defaultKeyValueStoreId"]).get_record("OUTPUT")["value"]
if out.get("sourceError"):  # a plan refusal ends SUCCEEDED with 0 rows — never read it as "nothing new"
    raise RuntimeError(out["sourceError"])
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["rank"], item["name"], item["pricePerNight"], item["whyThisHotel"])

🤖 Hotel search tool for AI agents (MCP)

{"mcpServers": {"apify": {"url": "https://mcp.apify.com/?actors=rich_minds/booking-hotel-finder-ai"}}}

Minimal agent input:

{"destination": "Lisbon", "checkIn": "30 days", "checkOut": "32 days", "maxQualified": 5, "maxDiscoveryChargeUsd": 0.5, "dedupeAcrossRuns": false}

Run outcomes — what your integration sees

👥 Who is it for?

🧠 How the AI works

• Grounded. One hotel per call; evidence and fine-print quotes must be exact substrings of the listing or they are dropped, and an invented number in whyThisHotel brings back the rule line.
• Your brief. briefMatchScore scores the hotel against travellerBrief; valueScore = 70 % rule + 30 % brief.
• Model. anthropic/claude-haiku-4.5 via Apify's model access, on the best candidates first; or your own key.
• Fallback. If the model fails, hotels keep the rule tier and the status says how many — free while your free hotels last, then at the basic price, never AI.

🔒 Data, compliance and limits

• Public Booking.com listing data only; host and trader contact data is ignored — no personal data.
• Prices change by the minute — confirm on Booking.com. The source run is capped by maxDiscoveryChargeUsd and 600 s; the $0.003 / $0.002 scrapers return no facilities or fine print. Stay within Booking.com's terms and GDPR.

❓ FAQ

How much will one run cost me? $0.50 of Booking Scraper for 100 properties ($0.30 with the $0.003 scraper) + 25 hotels × $0.004 (AI off) or ≈ $0.013 all in (AI on) here — first 10 free.

How long does a run take? Measured: the demo 2.8 s (AI off), 6.9 s (AI on); a live search adds the scraper's time. Apify's model access runs 8 calls at a time, your own key 2 — a free-tier key took 191 s for 4 hotels; waits stop after 2 min (60 s in all on your own key, OUTPUT.timing); worst case ≈ 17 min per 100 hotels. Pass timeoutSecs 1800.

Which Apify plan do I need? Any plan runs the demo and your own rows (list) or dataset; the live search needs a plan that runs Store Actors; the AI needs model access on your plan or your own key (llmProvider: "byok").

🧩 More Actors from the same developer

• Google Reviews Analysis AI — what guests say about a hotel or venue on Google Maps: complaints, praise, reputation
• Reddit Brand Monitoring — what travellers say about a hotel, brand or destination on Reddit, with sentiment

🆘 Support

Something missing or wrong? Open an issue on the Actor's page or message the developer through the Apify Console — requests from buyers are shipped first.

📝 Changelog

• 0.1.8 (2026-10-16) — One-click ready searches; one "Send alerts to" field for chat apps and webhooks; live AI verdicts on your first 5 hotels on any plan; price history in the client report; a shorter Store page that fits without being cut off.
• 0.1.5 (2026-10-04) — Bring your own data: rows in itemsList or a datasetId pick their mode, start no source Actor and are charged only for the analysis; unread fields are listed in the log.