Skip to content

Revenue Management

Which trends will shape hotel revenue management in 2027?

AI agents are moving from pilot to operation, costs are rising faster than rates and booking already starts inside AI. Six signals, what the big chains are already doing and where to lead in 2027.

By Rubén Sánchez Martín Published 11 min read

Silhouette of an executive on the phone at a hotel window overlooking the sea, with the text “2027, six signals for hotel revenue management”

By Rubén Sánchez Martín, co-founder and CEO of BEONx. Computer Engineer from the University of Salamanca, with executive education at Cambridge and Harvard. He has spent more than a decade helping hotel chains with revenue management technology.

Short answer

In 2027, hotel revenue management will stop being measured only in revenue and start being measured in profit. Six signals are pushing it. AI is already the norm in teams, agents are starting to execute complete tasks, every system in the stack brings its own AI, costs are growing faster than rates, distribution is concentrated in two groups and booking already starts inside assistants such as ChatGPT or Google. Hilton, Marriott, IHG, Accor, Wyndham and Meliá are already moving. To lead, a chain needs a single number, inventory and pricing ready for agents, decisions measured in profit and AI governance by design. That is what an Agentic RMS like BEONx Spark does.

Why is 2027 a different year for revenue management?

Hotel chains are closing 2026 with a paradox. Rates are going up, but profit is not keeping pace. In Europe, first-half RevPAR grew mainly on the back of ADR, according to CBRE, while staff, energy and brand and operator fees rose faster and squeezed gross operating profit. And for 2027, almost every forecast points to more moderate rate growth. If price stops doing the heavy lifting, profit will have to come from somewhere else.

At the same time, artificial intelligence is no longer an experiment. 90% of travel executives say their company already uses generative AI, and 80% plan to deploy agents at scale within the next three to five years, according to McKinsey. What has not changed yet is how a chain’s commercial strategy is executed. 2027 will be the year AI agents move from pilot to operation.

What are the six signals that are already here?

1. AI is the new normal

74% of professionals already use AI regularly, according to BCG. But very few companies have scaled it across the organisation. The chains that are doing so already measure the return. Meliá estimates that AI has freed up more than 100,000 hours of transactional and back-office work, and Hyatt now talks about moving from adoption to absorption, redesigning complete workflows and checking where the savings or revenue show up in the P&L.

2. Agents move from assisting to executing

30% of professionals already have agents built into their work, more than double a year ago (BCG). Hilton says it has left “pilot purgatory”, and Wyndham spent ten months piloting an LLM-based commercial platform covering revenue management, marketing and sales, already used by 115 revenue managers. Revenue management is, in fact, the first area where hotels have trusted AI with a pricing decision. The next step is to coordinate the whole stack with a single strategy. From executing to orchestrating.

3. The risk of a Frankenstein of AIs

Every vendor in the stack brings its own AI, with its own model, data, permissions and cost. Agent governance was the big topic at HITEC 2026, and in McKinsey’s report on agentic AI in travel, Marriott puts it in one line, “governance is critical”. Choice Hotels is building a semantic layer and MCP servers to create a “single governed door” for AI agents. Without an orchestrator, IT does not know which agents are acting, with what data or with what result.

Regulation adds to it. The EU AI Act requires telling guests when they are dealing with an AI and prohibits setting different prices based on a person’s socioeconomic profile, with fines of up to 3% of global turnover. Governing AI is no longer just good practice. It is an obligation.

4. Prices rise, profit evaporates

Energy prices are up 24% in 2026, the biggest rise in four years according to the World Bank. Diesel, close to 40% in Spain. Chemical raw materials for cleaning products and amenities, 70%. Rates are rising too, but not at the same pace, and staff shortages add to the pressure. That is why revenue forums such as HSMAI keep pushing the move from occupancy to TRevPAR, Net RevPAR and profit per available room. RevPAR is the output, not the goal.

5. Distribution is the biggest commercial cost

A €200 booking costs €40 through an OTA at 20% and around €14 as a direct booking. In Europe, two groups account for 85% of OTA bookings, intermediated sales are now close to 30% of the total and direct bookings have fallen six points since 2013, according to the HOTREC and HES-SO Valais study. The chains that hold up best are those that turn direct sales into strategy. Meliá already sells half of its centralised sales through its own channels, and its loyalty members generated more than 86% of its bookings in 2025. Every point of direct sales is profit.

6. Booking already starts inside AI

Since August 2026, Google lets US travellers book a hotel without leaving AI Mode, with Marriott, Hilton, IHG, Wyndham, Choice, Booking.com and Expedia as launch partners. Accor has had its own app inside ChatGPT since January, Hilton has taken its AI trip planner into ChatGPT, and its CIO warned at the NYU hotel investment conference that distribution could flip within a few years. IHG already measures each hotel’s visibility in AI engines with around thirty prompts per property. Demand is already automated. Many chains’ commercial strategy is not, yet.

An important nuance from Cornell’s Center for Hospitality Research. More than 60% of travellers cite accuracy as the main barrier to trusting AI, and luxury travellers prefer a person to finalise the trip. Trust is earned with accurate data and clear explanations, towards the guest too.

Demand is already automated. The chain’s commercial strategy is not, yet. That is the work of 2027.

What are the big chains already doing?

What the big hotel chains are already doing with AI (as reported by the companies in 2026)
ChainWhat it has doneWhich signal it confirms
HiltonAI planner on its website, app and ChatGPT. Its messaging assistant handles 40–45% of guest messagesBooking starts inside AI
MarriottGuest trip-planning assistant, direct booking from Google AI Mode and a ChatGPT appBooking starts inside AI
IHGAI-ready hotel content and per-hotel visibility tracking in AI search enginesHotels must be readable by agents
AccorIts own app in ChatGPT that sends bookings to its websiteThe battle for direct comes earlier
WyndhamLLM-based commercial platform for revenue, marketing and sales, used by 115 revenue managersAgents reach revenue management
ChoiceSemantic layer and MCP servers as a "single governed door" for agentsAI governance built into the architecture
HyattFrom AI use to redesigning complete workflows measured in the P&LAI is measured in profit
MeliáMore than 100,000 hours freed by AI and half of centralised sales through its own channelsEfficiency and direct sales as levers

Where does a hotel chain need to lead in 2027?

This year’s reports agree on one point. Most travel organisations are experimenting with agents, but very few have scaled them. 89% of hotel organisations still do not have an agent able to complete bookings and price inventory in real time, according to a study by Aven Hospitality and h2c published by Skift. That gap is the advantage of whoever moves first. These are the five fronts where it pays to be ahead.

  1. One number and one brain. In many chains, revenue, sales, operations and finance each work with their own forecast, and none of them match. Leading means a single forecast per hotel, recalculated continuously, that every department and every agent works from.
  2. Inventory, pricing and content ready for agents. If an AI assistant cannot query, compare and book your inventory with reliable data, it will recommend another hotel. That takes consistent pricing across channels, machine-readable content and open connections through APIs and MCP.
  3. Decisions measured in profit. Choosing what each hotel maximises, from RevPAR to Net RevPAR, and getting ready for TRevPAR and GOPPAR, with each channel’s cost inside every price.
  4. AI governance by design. Role-based permissions, limits the agent cannot cross, a log of every action and compliance with the EU AI Act, with no prices based on a person’s profile.
  5. A team that orchestrates. The revenue manager stops loading prices and building reports and moves to designing strategy, challenging the agents and deciding how much autonomy to give them.

What does each signal mean for a hotel chain?

Six signals for 2027 and what to ask of your revenue management system
SignalWhat it means for a chainWhat to ask of the RMS
AI is the normTeams already using AI on their ownA single AI platform, with role-based permissions
Agents that executeComplete tasks that can be delegatedAgents that act within the team's limits
Frankenstein of AIsMany AIs with no common governanceOne orchestrator that logs every action
Rising costsRevenue that does not reach profitChoosing each hotel's KPI, from RevPAR to profit
Concentrated distributionCommissions that erode EBITDANet RevPAR, with each channel's cost in every price
Booking inside AILate decisions and a hotel AI cannot seeContinuous recalculation, consistent pricing and APIs and MCP

What should the board decide before 2027?

Giving an agent autonomy is a leadership decision, not an IT one. Before signing off the budget, the board should leave the room with three decisions made.

  1. Which objective. Which KPI each hotel pursues, from RevPAR to Net RevPAR, and how it is measured.
  2. How much autonomy. What the agent does on its own, what needs approval and which controls IT needs to see working.
  3. How to test it. A representative group of hotels, an agreed baseline and a business case in three scenarios.

BEONx Spark is the first Agentic RMS. Its AI agents execute a hotel chain’s commercial strategy 24/7, across its whole portfolio, within the limits the team sets and explaining every decision.

  • One Brain, one number. A single brain with a multi-level forecast per hotel, 95% accurate and recalculated continuously, that orchestrates the agents and the rest of the chain’s stack instead of a Frankenstein of AIs.
  • One agent per objective. The chain chooses what to maximise. RevPAR Agent and Net RevPAR Agent today, TRevPAR and GOPPAR coming soon.
  • Open to the chain’s agents. Two-way integration with PMS, channel manager and CRS, and APIs and MCP to connect the rest of the stack and the AIs the chain already uses.
  • Harness. Role-based permissions, limits, approvals and a log of every action, with each chain’s own data and in line with European regulation. An agent can only do what its user allows.
  • From revenue to profit. The Net RevPAR Agent puts each channel’s cost into every price, across every segment, tour operators included.
  • You talk to it. The team asks in plain language, by voice too, and gets every decision with its reason. The language model is the mouth, not the brain. Prices are calculated by BEONx’s revenue algorithms.

Frequently asked questions

AI as the norm in teams, agents that execute complete tasks, the risk of many AIs without governance, costs growing faster than rates, distribution concentrated in two groups and booking that already starts inside assistants such as ChatGPT or Google. All of them push revenue to be measured in profit, not just in revenue.

What are the big hotel chains doing with AI?

Hilton, Marriott, IHG and Accor are already present inside ChatGPT or Google AI Mode. Wyndham uses an LLM-based commercial platform for revenue, marketing and sales, Choice is building a single governed door for AI agents and Meliá has freed more than 100,000 hours of work with AI.

What is an Agentic RMS?

A revenue management system whose AI agents execute the chain’s commercial strategy and orchestrate its technology stack, within the rules the team sets and explaining every decision. BEONx Spark is the first Agentic RMS.

How does the EU AI Act affect revenue management?

It requires transparency when guests interact with an AI and prohibits setting different prices based on a person’s socioeconomic profile. For a chain, it means choosing systems with permissions, a log of every decision and prices based on demand and the market, not on the guest’s profile.

Why is RevPAR no longer enough?

Because it does not see what it costs to sell each room. With distribution and operating costs rising, two bookings at the same price do not leave the same money in the hotel. Net RevPAR deducts each channel’s cost and brings the commercial decision closer to EBITDA.

Will AI agents replace the revenue manager?

No. Agents take over what was never revenue management, such as building reports, copying data between systems or changing prices cell by cell. The revenue manager gains time to design strategy and moves from operator to orchestrator.

Sources:

Free download: “The Agentic RMS Guide for Hotel Chains”

2027 trends, how the agents work, how they are governed and where to start. Written for executive committees and revenue teams.

Download “The Agentic RMS Guide for Hotel Chains” → · Talk to an expert →

Keep reading: Why do hotels miss profit with manual revenue tools?

About BEONx. BEONx was founded in Salamanca in 2012 to help hotels maximise their profitability. In 2026 it launched BEONx Spark, the first Agentic RMS: AI agents that execute a hotel chain’s commercial strategy 24/7, across its whole portfolio, within the limits the team sets and explaining every decision.

Related articles

See all →

Unlock your potential

Speak with one of our dedicated hotel experts and get started on your journey to sustainable profitability.

Other Articles