Hotel Distribution's Data Isn't Granular Enough for the AI Agents About to Search It
Consultant Martin Soler argues conversational AI could finally make attribute-based selling real in hotel distribution — but only if hotels fix property data that still records amenities as binary yes/no fields.
Martin Soler, Partner at Soler & Associates, argues the 30-year-old battle for hotel distribution’s customer interface is starting over — this time with AI doing the searching instead of a human scrolling through 200 properties. Soler’s example: a guest could simply describe “a quiet boutique hotel in Rome, walking distance from good restaurants, with a pool suitable for toddlers” and let an AI agent do the matching, rather than filtering OTA results by star rating and price.
His technical argument is the more durable one. Today’s distribution infrastructure — OTA, GDS, PMS, CRS systems — is built around basic availability, rates, and inventory (ARI) data that works fine when a human does the filtering, but breaks when an AI agent has to match nuanced requests against it: most properties still record amenities as binary fields like “Pool: Yes/No” rather than the descriptive detail — toddler-safe depth, heated, indoor — an AI agent actually needs. Attribute-based selling has been discussed in hospitality for years without real adoption; Soler suggests conversational AI search may finally force it, calling it “attribute-based search on steroids.” He flags the structural catch: upgrading that data infrastructure doesn’t generate immediate revenue, so whether OTAs and their investors commit to the overhaul is an open question with direct consequences for how hotels get discovered next.
The data-granularity problem isn’t new to hospitality-tech observers. Mirai’s Pablo Delgado made a related argument in a September 8 Hospitality Net piece, estimating that only about 30% of the questions travelers actually ask can be answered from a hotel’s own website today, with the rest depending on operational knowledge that simply isn’t published anywhere a machine can read it. Delgado’s fix runs through MCP endpoints and machine-readable “agent-ready” hotel data; Soler’s runs through richer ARI attributes. Different vocabulary, same underlying diagnosis: hotel data was built for humans, and the AI-agent era is exposing how much of it was never actually structured at all.
Source: Hospitality Net Auto-generated brief — verified before publishing.