At Food Hotel Tech 2026 in Paris, Aude Naveilhan of Mirai and Killian Defois of SiteMinder explored how artificial intelligence can improve hotel profitability beyond the topline. Hoteliers too often conflate revenue with profit: the panel opening this year’s show reminded attendees that RevPAR, the sector’s favoured metric, is a blunt KPI that ignores commissions, discounts and a host of other costs. Naveilhan, Mirai’s country manager, argued for steering by the net average rate, a measure that fully accounts for distribution, marketing and financial costs. That approach, the speakers said, reveals cases where an extra sale ultimately destroys value, “more often than you think”, as Defois, SiteMinder’s France country manager, put it, because of threshold effects such as opening a floor or calling in additional staff.
Profitability is also shaped by commissions paid to online travel agencies, far beyond the headline percentage taken per booking. Defois noted that customers booking direct are more inclined to choose premium room types, extend their stays and spend more on ancillaries. “Direct bookings generate on average 65% more revenue than on OTAs, excluding commission,” he said. Loyalty therefore becomes a margin lever, with fewer cancellations and higher baskets.
Beyond ideology
To improve a hotel’s margins, the goal is to loosen dependence on OTAs. AI is redrawing the visibility game. A growing share of travellers plan trips with conversational assistants, a trend that is even more pronounced among younger guests. To appear in large language model (LLM) answers, hotels need to structure their content, ensure information is reliable and, ideally, expose it via an MCP server to limit agent errors and answer long, contextual queries precisely. This shift towards conversational discovery requires hotels to adapt content, schemas and journeys.
Is the future of hotels to be found far from OTAs? The speakers preferred to talk about arbitrage. “It’s a false debate,” Defois said. “They are all acquisition channels. The question is what the return looks like. OTAs have become the first place travellers search, ahead of search engines. But 20% of travellers who start on OTAs finish on direct, and that share is growing. You can treat OTAs as a marketing channel that may cost less than pure Google Ads investment or maintaining a CRM. We need to move past ideology and into pure return logic: use OTAs to feed direct, then lean on loyalty.” The operational aim, then, is to calculate the net average price by channel and allocate budgets to the flows with the best yield.
From operator to strategist
AI also plays a role in pricing, the panellists said, by making hotels more responsive. An event announcement can trigger demand spikes that require rapid adjustments, and opportunities are missed when rates move too slowly. AI helps listen to the market in real time, digest large data volumes and propose pricing scenarios.
AI will not make OTAs disappear any time soon — the platforms are investing in AI themselves. But there is room to manoeuvre. Independent hotels can regain ground by measuring real profitability by channel, building a data/AI infrastructure (structured data, an MCP server) and shifting from an operator’s role to that of a strategist across channel mix and pricing. Loyalty and pricing agility then become critical levers.
_w=800_h=446_pjpg.jpg?v=20230522122229)
