AI just turned your channel manager from back-office plumbing into who controls your guest.

TLDR

• Expedia Group's Censuswide survey of 1,500 hotel decision-makers across six markets, published June 23, 2026, found fully connected hotels are 31 percentage points more likely to report improved occupancy, ADR, or RevPAR than hotels with only basic connectivity.
• A RateHawk and Phocuswright report, published on Hospitality.today on June 24, 2026, argues AI agents can only book machine-readable inventory, and that work is happening inside wholesalers and channel managers, not at the property level.
• The behavior shift: the platforms that control your connectivity layer now decide which AI assistants can surface your rooms, and most hotel contracts written before 2024 never addressed that question.

Expedia Group's own research, a Censuswide survey of 1,500 hotel decision-makers across the US, Canada, Mexico, France, Italy, and South Korea published June 23, 2026, reframes hotel connectivity from back-office infrastructure into the layer that decides which AI-powered channels can reach your rooms at all. A separate report from RateHawk and Phocuswright, published June 24, 2026, makes the mechanism explicit: AI agents can only transact on inventory that's machine-readable, and today that translation work happens inside wholesalers and channel managers, not at the hotel itself. For independent hotels, the question is no longer "are we connected." It's "who controls the connection, and what can they do with it without asking us."

Expedia's own research shows fully connected hotels report a 31-point revenue advantage over basic connectivity.

Expedia Group's Censuswide survey, conducted March 26 to April 7, 2026, found 81% of fully connected hotels report improved occupancy, ADR, or RevPAR, against 52% of hotels with only basic connectivity, a 31-point gap. Chris Hodges, Expedia Group's Vice President of Global Connectivity & Partner Solutions, framed the shift directly: "As booking channels proliferate, travel distribution is increasingly complex." The same survey found 95% of fully connected properties confident their rate changes propagate across channels within 15 minutes, against a much lower share for basic connectivity. Expedia isn't just measuring connectivity as a cost-saving convenience anymore. It's measuring it as the precondition for whether a property can compete at all.

Take a 35-room hotel still on extranet-only connectivity, manually re-entering rates across Booking.com, Expedia, and a handful of smaller channels. A rate change that takes that property 45 minutes to push out manually would propagate in under 15 minutes for a fully API-connected competitor down the street, a gap that compounds every time either property adjusts for a citywide event or a last-minute cancellation wave. The manual property is still setting the same rate. The connected one is setting it three times a day without anyone touching a keyboard.

AI agents can only book inventory that's machine-readable, and hotels aren't the ones doing that translation work.

The RateHawk and Phocuswright report puts the mechanism in blunt terms: "a hotel's inventory has to be readable by a machine before an AI agent can book it." Most hotel inventory, rate plans written in free text, cancellation terms buried in prose, was built for a human revenue manager or a booking-engine template to interpret, not for an autonomous system to parse. Critically, the report notes this translation work is happening "one layer removed from the property, inside the wholesalers, channel managers, and aggregators," which means hotels are dependent on those intermediaries to make their own inventory legible to AI, rather than doing it themselves. RateHawk's own connected inventory has doubled to 250,000 properties as it builds out that translation layer. The hotel owns the room. The channel manager owns whether an AI agent can understand what that room costs and whether it's available.

A 40-room independent hotel with a rate plan that reads "Non-refundable, book by 6pm the day prior, subject to seasonal blackout dates, ask front desk for details" is unreadable to an AI agent trying to quote a price in real time. The same rate plan, restructured into a channel manager's standardized rate-and-restriction fields, three date ranges, a cancellation cutoff, a blackout calendar, becomes something an AI agent can parse and quote in milliseconds. The rate didn't change. Whether an AI agent can sell it did.

This isn't a new dynamic. Platforms have used connectivity and payment control to shape distribution before.

Platforms controlling the infrastructure layer to shape distribution has precedent. Google Hotel Ads' commission-based bidding once made metasearch accessible to smaller properties with real-time rate feeds; that model was eliminated in 2024 and 2025 under pressure from the EU's Digital Markets Act and replaced by cost-per-click and target-ROAS bidding that favors bigger advertisers. In each case, the platform that controlled the technical layer, live-rate feeds then, machine-readable inventory now, decided which properties got distribution reach and which didn't. Hotels didn't lose visibility because their rooms got worse. They lost it because the platform changed what the technical layer required, without asking.

A 25-room hotel that invested in real-time rate connectivity ahead of Google's metasearch bidding change kept its visibility when the model shifted in 2024 and 2025, while a comparable property still on static-rate feeds saw its metasearch placement drop regardless of how competitive its actual rate was. The same pattern is now setting up around AI-agent connectivity: properties that get their inventory into machine-readable form early are positioned the way the real-time-rate hotels were before 2024. Being ready for the infrastructure requirement before it's enforced is worth more than being ready to compete on rate.

The framework: infrastructure-ready inventory versus infrastructure-dependent inventory.

Independent hotels can sort their own connectivity into two categories. Infrastructure-ready inventory is machine-readable today, structured rate plans, restrictions, and availability a system can parse without human interpretation, whether that structuring happens through the hotel's own PMS or a channel manager acting on its behalf. Infrastructure-dependent inventory still relies on a human, a revenue manager, a channel-manager account rep, to translate free-text rate plans into something a platform's API can use, which means every new AI-agent integration waits on that person's availability. Fully connected hotels in Expedia's survey are disproportionately in the first category. Extranet-only hotels are entirely in the second.

Branded chains negotiate connectivity terms centrally. Independent hotels take whatever their channel manager already agreed to.

A 200-room branded property with a revenue-management team can negotiate AI-connectivity terms directly with Expedia Group or a channel-manager provider, turning the question into a contract term it controls. A 25-to-40-room independent hotel using a standard channel-manager agreement has no equivalent negotiation. It takes the connectivity package the platform offers or loses distribution reach entirely. US independent hotels are somewhat more likely to already run a modern cloud PMS with API-native connectivity, while many European independent properties, especially older family-run hotels, are still further back on the extranet-only end of Expedia's own survey categories, per the six-market breakdown in Expedia's research.

Three things to do this quarter: audit your API access, check your AI opt-out rights, and decide your position on AI search.

• Audit which platforms have API access to your inventory today: check the connection type for each OTA and channel manager, since extranet-only connections mean you're already invisible to AI-powered search on that platform.
• Ask your channel manager whether you can opt out of specific AI integrations while staying API-connected: most contracts written before 2024 don't address AI placement at all, which means silence gets read as consent.
• Decide your own position on AI search placement before the platforms decide it for you: weigh the extra distribution reach against the loss of rate visibility and guest attribution, and say where you stand to your channel-manager vendor.