Skip to main content

How OTA Ranking Algorithms Actually Work

The signals Booking.com, Expedia, and Airbnb actually use to sort hotels, and what moves them

PillarOTA ranking algorithmAnya CortezReviewed Apr 18, 2026

How OTA Ranking Algorithms Actually Work

Sources: Booking.com Partner Hub and Expedia Partner Central pages (both verified April 2026 via Playwright), peer-reviewed academic research on hotel meta-search re-ranking, industry studies, and field observation from Booking.com search results in Barcelona and Albufeira. Last reviewed: 2026-04-18.

Key takeaways

Booking.com officially names five ranking signals. Expedia names twelve (four Offer Strength factors plus eight Guest Experience factors), in published priority order, and admits that commission level is also a ranking input 12. Both pages are public. Both are also the kind of partner-hub documentation most hoteliers skim once at onboarding and never reread.

The mistake that costs the most is not "low review score." It is misreading the algorithm as a single dial. Every signal falls into one of four categories: revenue, listing quality, guest experience, or paid participation. Optimizing one category while ignoring the others leaves a hotel ranking below comparable properties with more balanced profiles. Field observation across the top-popularity Barcelona search results bears this out: two 8.9-scored properties ranked below several 8.1s, suggesting that review score alone does not determine sort position 3.

This article reproduces both OTAs' verified ranking signals, organizes them into the four-category framework, and walks through how to diagnose where a property is losing rank. Sections 5 and 6 give the workflow and the audit checklist.

Why it moves bookings

Top ten on Expedia means 62 percent of clicks. That's Expedia's number, from their own June-December 2023 internal data 1. The 50-plus properties below the top ten split the other 38 percent. Page two to page one is not incremental. It's a cliff.

The Booking.com Partner Hub is more abstract about position economics, but the algorithm's stated goal is the same: sort properties by expected value to the platform. The five officially-named signals are conversion rate, average daily rate, cancellation rate over the last 90 days, property page score, and guest review score 2. These are the things Booking explicitly tells partners to focus on. They are not the only things the algorithm uses, but they are the things Booking is willing to confirm.

Expedia is more transparent. Its blog post "Decoding our algorithm" lists Offer Strength factors (room availability and inventory, rate competitiveness, content completeness, photo quality and quantity) and Guest Experience factors (preventable relocations, property condition and facilities, preventable cancellations, cleanliness, check-in ratings, preventable refunds, amenities, staff and service reviews) in priority order 1. Booking wants to show the listing that will book. Higher commission tilts it further. That's the Expedia admission, in plain text: "we consider how much we're paid when a traveler stays at your property, which includes commissions from accommodation and compensation on bookings." Most ranking-factor articles skip this admission.

The honest caveat: the algorithm can change overnight without warning, and hoteliers don't control commission tier, paid-participation pricing, or whatever pCTR/pCVR weights Booking decides to reshuffle this quarter. What you do control is the list of signals named on both OTAs' partner pages. That list is this article.

What "great" looks like

Every signal Booking and Expedia name falls into one of three operational categories, plus a fourth that sits as a paid overlay on top of the algorithm rather than inside it. The taxonomy is ours, derived from both OTAs' official lists.

Category 1: Revenue signals. What you generate for the OTA. Conversion rate (Booking #1), average daily rate (Booking #2), rate competitiveness (Expedia Offer Strength #2), and commission level (Expedia explicit). Higher conversion plus higher commission per booking equals more value to the OTA per impression. The algorithm rewards both.

Category 2: Listing quality signals. How complete and attractive your property page is. Property page score, which Booking explicitly defines as content, pictures, descriptions, and amenities (Booking #4). Room availability and inventory (Expedia Offer Strength #1). Content completeness (Expedia Offer Strength #3). Photo quality and quantity (Expedia Offer Strength #4). Sparse listings under-convert and the algorithm notices.

Category 3: Guest experience signals. How well you deliver after the booking. Guest review score (Booking #5), preventable relocations (Expedia Guest Experience #1, top-weighted), property condition and facilities (Expedia #2), preventable cancellations (Expedia #3), cleanliness (Expedia #4), check-in ratings (Expedia #5), preventable refunds (Expedia #6), amenities (Expedia #7), staff and service (Expedia #8), and Booking's 90-day cancellation rate (Booking #3). Note that Expedia weighs preventable relocations above cleanliness. Most hoteliers reverse those priorities.

Plus a fourth, paid overlay: Booking.com Preferred Partner status, Genius discounts, mobile-rate participation, Expedia Accelerator and TravelAds. These move position without improving the underlying property profile. They are a margin tax to compensate for, not replace, weak Categories 1-3. Approximate cost bands: Accelerator runs as a sliding additional commission (typically 1-10 percent on top of base); Genius is a 10 percent rate discount. A 45-room independent can apply for Preferred Partner once Booking signals consistent performance, but the program meaningfully favors high-volume properties. We treat the paid overlay as a sidecar in this article and revisit it in the Pricing cluster.

Officially-unconfirmed signals practitioners cite. Response time across Booking, Airbnb, and VRBO has become a heavily-discussed ranking input in 2026 partner-community channels 4, and Booking's Preferred Partner program weights average response time. Neither is on the official five-factor list, but if your account manager says it matters, treat it as real.

Airbnb is out of scope here. They do not publish their ranking signals the way Booking and Expedia do, so we couldn't apply the same primary-source treatment. A sibling article will cover Airbnb when their algorithm gets a public surface worth citing.

The case study Expedia itself published. The Marker Key West Harbor Resort improved its guest experience score by more than 20 points and grew revenue 11 percent year-over-year in the first half of 2023 by acting on the personalized recommendations in Partner Central 1. This is Expedia citing one of its own partners. The mechanism: not advertising spend, not a price cut. Operational fixes traceable back to Category 3 signals.

Common failure modes

Optimizing one category and skipping the others. Fix conversion but skip content score, and you rank below hotels that covered both. The algorithm rewards spread, not depth.

Ignoring preventable cancellations and relocations. Both Expedia and Booking weight these heavily, and most hoteliers do not treat them as ranking signals. They treat them as fulfillment problems. The Expedia priority order is unambiguous: preventable relocations sit above cleanliness, amenities, and staff. A front desk that over-sells by three rooms on a Friday is, in algorithmic terms, sabotaging its own visibility.

Assuming review score is the dominant signal. It is one of five Booking signals, not the master signal. We checked the Barcelona top-popularity sort on April 18, 2026: two 8.9-scored properties (Hostal Girona and Tembo Barcelona) did not lead the grid. They sat behind several 8.1- and 8.4-rated properties. The likely explanation is conversion rate, commission tier, or paid-participation, but the visible takeaway is that review score does not linearly map to position.

Betting on paid placement to compensate for weak fundamentals. Accelerator and TravelAds boost visibility, but if the underlying Offer Strength and Guest Experience scores are weak, the paid traffic converts poorly, which then hurts organic rank. Paid is a multiplier on a healthy listing, not a substitute for one.

Citing 2022 ranking-factor articles in 2026. Expedia migrated from a single quality-score model to the two-factor Offer Strength plus Guest Experience model mid-2022. Booking.com's Partner Hub ranking page is dated 2023 and the actual ML almost certainly evolves between annual updates. Articles that quote pre-2023 algorithm explanations are quoting a model the OTA no longer uses.

How to diagnose your rank position

Pull each OTA's official scores, map each weak score to the framework, fix the highest-weighted weak signal first, then re-check in 30 to 90 days.

  1. Pull your Booking.com Ranking Dashboard. In the Extranet, go to Analytics, then Ranking dashboard. Review the 30-, 90-, and 365-day trends for Search Result Views, Click-Through Rate, Property Page Views, Bookings, and Search Results Score (your performance relative to your market) 2.
  2. Pull your Expedia Offer Strength and Guest Experience scores. In Partner Central, go to the visibility performance page. Each of the twelve factors has a personalized recommendation attached.
  3. Map each low-scoring factor to the framework above. A property might be revenue-strong but listing-quality-weak (a common urban-hotel pattern), or guest-experience-strong but paid-participation-weak (a common boutique pattern).
  4. Fix the highest-weighted weak signal first. Expedia explicitly orders Guest Experience factors by impact: preventable relocations above property condition above preventable cancellations above cleanliness. Fix in published order. For practical guidance on the operational signals (preventable relocations comes from over-selling and slow turn-arounds; sibling articles cover negative-review responses and photo quality at depth), see the Related articles list below.
  5. Re-check in 30 days. The ML averages signals across windows. Daily checks are noise.
  6. Wait 90 days before judging if a change worked. Cancellation rates roll over a 90-day window on Booking. Review-score trends need at least one full booking-and-review cycle to surface.

Self-audit checklist

Run this on your own property, without our product:

  • I have viewed my Booking.com Ranking Dashboard in the last 30 days
  • I have viewed my Expedia Offer Strength and Guest Experience scores in the last 30 days
  • I can name my Conversion rate, ADR, and 90-day cancellation rate from memory
  • I know my current Search Results Score (Booking) versus my market
  • I know my preventable relocations and preventable cancellations count for the last 12 months
  • My listing has photos, descriptions, and an amenity list filled to at least the market median (Property page score input)
  • My review response rate sits between 30 and 50 percent overall, with 100 percent on negative reviews
  • I know which paid-participation programs I am enrolled in (Preferred Partner, Genius, Accelerator, TravelAds) and the cumulative commission cost
  • I have not relied on a pre-2023 ranking-factor article in the last six months

How OTALift surfaces this

Our listing-audit and review-ongoing reports map onto the four-category framework. PhotoQualityValidator (Category 2), EngagementValidator (Category 3 review-response slice), and rate-parity checks (Category 1 commission-tier slice) cover specific signals; the report aggregates them into a property-level scorecard.

The research surfaces three additions worth implementing: a ResponseTimeValidator (the response-time signal practitioners cite but neither OTA officially confirms; worth tracking even if its weight is uncertain), a content-completeness-versus-market validator (Expedia's Content Score uses comparable-property benchmarking; today our check is binary, not relative), and a cancellation-rate signal at the channel level (Booking explicitly weights 90-day cancellation rate; today our reports track conversion but not the cancellation slice). All three are tracked in the internal report-improvements backlog as direct outputs of this article's research.

Related articles

Sources and methodology

Authored by Anya Cortez · Reviewed by Tim Anastasiou · Last reviewed: 2026-04-18

Anya Cortez is OTALift's hospitality researcher and writes The Labs.

Footnotes

  1. Expedia Group Blog, "Decoding our algorithm: A guide to boosting your hotel's visibility." Verified 2026-04-18 via Playwright. Source for: 62% top-10 click share (Expedia Group internal data, June-December 2023), Offer Strength and Guest Experience factor lists in priority order, the commission-as-ranking-factor admission, and The Marker Key West Harbor Resort case study (+20 guest experience score points, +11% revenue YoY H1 2023). https://partner.expediagroup.com/en-us/resources/blog/travel-marketplace-visibility-guide ↩ ↩2 ↩3 ↩4

  2. Booking.com Partner Hub, "Search results, ranking, and visibility." Updated 2023, accessed 2026-04-18 via Playwright. Source for: the five officially-named ranking signals (Conversion, ADR, 90-day Cancellations, Property page score, Guest Review Score) and the ranking-dashboard metric set. https://partner.booking.com/en-us/help/growing-your-business/analytics-reports/search-results-ranking-and-visibility ↩ ↩2 ↩3

  3. OTALift field observation, Booking.com Barcelona search results captured 2026-04-18 via Playwright. Two 8.9-scored properties (Hostal Girona, Tembo Barcelona) sat behind several lower-scored properties in popularity-sorted order. Screenshot at docs/labs/articles/listings/perfect-exterior-photo/media/booking-barcelona-search-results.png. ↩

  4. Aeve AI, "How Response Time Impacts Airbnb, VRBO and Booking.com Rankings 2026." Practitioner-cited finding that round-the-clock response time is weighted heavily in Booking.com Preferred Partner status. Practitioner-inferred; not on Booking's official 5-factor list. https://www.aeve.ai/blog/airbnb-vrbo-response-time-listing-ranking-2026 ↩

Related articles

Want OTALift to apply this to your property?

Every recommendation in our reports links back to one of these articles.

Book audit