10-Room Historic Waterfront Inn
Independent inns lose direct-booking share to OTAs not because OTAs are better at hospitality, but because OTAs are better at being machine-readable. Booking.com publishes structured data for every property — `LodgingBusiness`, `Hotel`, `Room`, `Offer`, `Review` — that ChatGPT, Perplexity, and Gemini extract at scale. An independent inn with no `LodgingBusiness` schema, no `Room` typing, and no machine-readable rate card is structurally invisible when a guest asks 'historic boutique inn waterfront South Carolina.' The audit's job is to make the property's own substrate as legible to the engines as Booking.com's, then deploy a property-specific AI concierge to capture the late-night decision window OTAs currently own.
A 233-year-old waterfront inn, ten rooms, 4.8-star reputation, taken over by new innkeepers in August 2023. Twenty months in. Capital work done, systems settled, reputation strong. The question had shifted from 'can we run this property' to 'where is the operational leverage.'
- OTA commission was quietly leaking $28K–$55K annually. Boutique inns in the 8–15-room tier in the Southeast typically see 35–55% of bookings arrive through Booking.com, Expedia, and VRBO — at 15–25% commission. Direct-booking share was below the 25–35% benchmark for a well-run inn at that scale.
- The late-night booking window was dark. Between 9pm and 1am — when vacation-decision intent actually peaks, the end-of-workday planning window — the site was silent. Booking.com's 24/7 chat was answering the same questions and getting the booking.
- Two specific moments were losing direct-booking intent: the pre-booking question (parking, breakfast, dog policy, room-view specifics, whether the carriage house differs from the main house) — currently emailed to concierge@ and answered whenever the innkeepers could get to it; and the late-night window specifically — when the humans are asleep and the OTAs aren't.
- Ten rooms × handful of repeat questions = an AI concierge use case. 60–70% of inquiries were repeats of the same dozen questions. A property-specific AI trained on real room descriptions, policies, and FAQ could handle those in-moment, 24 hours a day, and escalate the genuine conversations to the innkeepers with full context attached.
At the inn's scale, every 5-percentage-point shift from OTA to direct is worth approximately $8,000–$15,000 in annual margin. A move from a conservative 25% direct baseline to a 35% direct share is $16K–$30K annually. The AI concierge engagement pays itself back inside 6–10 months.
A 14-day engagement: property-specific AI concierge trained on actual room descriptions + policies + FAQ, integrated with the PMS for live availability, live on-site as a chat widget with same-page booking capture. Handles repeat questions in-moment; escalates real conversations with full context. Not a chatbot with scripted answers — a narrow, property-specific assistant.
A 14-page audit with the OTA commission math modeled against their actual scale, the specific two moments where direct-booking intent is currently lost, and a 30-60-90 day rollout plan for the AI concierge with expected direct-booking lift per phase.
The named JSON-LD types deployed in the engagement. Each one is the layer the answer engines extract before they read the prose — pick one that's missing on your own site and you have your first sprint.
LodgingBusiness— root property type with `address`, `geo`, `numberOfRooms`, `starRating`, `petsAllowed`, and `checkinTime` / `checkoutTime`Hotel— subtype of LodgingBusiness — historic-property classification with `yearBuilt` and historic-register referencesRoom— per-room schema with `bed`, `occupancy`, `amenityFeature`, `floorSize`, and `view` — so engines answer 'inn rooms with water view in [town]' correctlyOffer— rate-card schema with `priceCurrency`, `availability`, and seasonal `priceSpecification` — exposes direct-booking pricing the OTAs can't undercut onFAQPage— the dozen repeat questions (parking, breakfast, dog policy, room differences, cancellation, late check-in) deployed as structured FAQReview + AggregateRating— 4.8-star reputation across booking platforms aggregated with attributionGeoCoordinates— precise lat/long for waterfront proximity — answers location-specific intent queriesTouristAttraction— the property itself as a 233-year-old historic destination, separate from the LodgingBusiness layerBreadcrumbList— site hierarchy across rooms / rates / experience / contact tree
Six engines, the same query set, the same time window. We record what gets cited verbatim — which engine, which firm, which quote. The output is the evidence behind every finding in the audit dossier.
- · ChatGPT 4o
- · Claude Sonnet
- · Perplexity AI
- · Google Gemini
- · Microsoft Copilot
- · Grok
- “historic boutique inn waterfront South Carolina”
- “10-room boutique hotel coast SC”
- “alternative to Booking.com for [coastal SC town]”
- “direct booking historic inn Lowcountry SC”
- “pet-friendly historic inn coastal South Carolina”
- “best small inn with water view South Carolina”
- “boutique stay near [coastal SC town]”
Want the methodology behind every claim above? Here are the source pages.
Every audit follows the same shape; the diagnosis is what changes. See the methodology in the sample audit deliverables, or compare verticals across all four shipped case studies.