Hotels are increasingly using artificial intelligence (AI), but manual work continues to weigh teams down.
A new report from H2c found that 69% of hotels still have staff manually entering guest preference data, compared to 19% using AI-driven extractions from guest interactions.
Moreover, the State of Distribution 2026 report from New York University, RateGain and HEDNA found that more than 80% of commercial teams still spend one to two days a week producing and analyzing reports manually.
The distribution report also revealed more than half of hotels are using or procuring generative AI, but fewer than one in 10 report reductions in manual work greater than 30%.
According to Amanda Moore, VP of digital experience at Preferred Travel Group, much of initial adoption has been task based, and the sector's longstanding fragmentation issues stand in the way of true transformation.
"Layering AI on top of disconnected data does not eliminate that underlying complexity," Moore said.
When asked about how data quality affects efficiency, Ksenia Tarasova, CRM and brand loyalty manager at Penta Hotels, also noted that "AI is only as useful as the data and processes behind it."
"If guest profiles are duplicated, preferences are inconsistent, consent isn’t properly structured or data is spread across several systems, putting AI on top doesn’t solve that problem—it can actually amplify it," she said.
Legacy systems are another component, but Tarasova said that connectivity is the larger issue.
"A mature system with good APIs and clean data can be much more valuable than a brand new AI platform sitting in another silo," she said. "I also think we’re currently seeing a lot of AI added to individual tools rather than AI working across the full guest journey … We are not 100% ready for AI as an industry."
The midsize 'sweet spot'
Both Moore and Tarasova highlighted AI's impact on midsize hotel groups—defined as chains with less than 100 properties and a presence in a few countries—with the State of Distribution report identifying these chains as a "potential commercial 'sweet spot.'"
"Midsize chains represent the most balanced model in 2026. They have enough scale to support specialized talent and automation while remaining agile enough to avoid the legacy silos that hinder global brands," the report reads.
Midsize hotel chains had the highest adoption of preset automation when compared with large chains and independent hotels, and they were also the most likely to report significant labor reductions, with about 14% reporting a decrease of more than 30%.
Additionally, their AI search accounted for 6% of their reservations, which was the highest of the three hotel-size categories.
"There seems to be a point where an organization is large enough to invest properly in technology and specialist teams but still small enough to avoid some of the complexity and silos that come with very large technology ecosystems," Tarasova said.
Moore suggested this may indicate that success is not necessarily tied to spend.
"It is about connecting data, people and technology well enough to act quickly," she said.
OTAs, AI and the direct relationship
The State of Distribution report also found that hotels' core direct-booking capabilities are "nearly universal" for hotels, but online travel agencies (OTAs) still have a larger share of bookings.
In 2026, OTAs account for 34% of bookings, while hotel websites and mobile apps, the largest direct-digital source, account for 18%.
According to Moore, hotels have addressed infrastructure with strong websites and booking engines, but demand is the bigger challenge.
"OTAs remain incredibly effective at capturing travelers during the discovery and consideration stages, and independent hotels, in particular, cannot simply outspend them," she said.
To win, hotels need to give travelers a reason to book direct.
"That means differentiated content and storytelling, compelling member or direct-booking value, strong search and metasearch visibility, frictionless booking experiences, personalization and better use of first-party data and CRM after a guest enters the ecosystem," Moore said.
“For independent hotels, especially, differentiation is the advantage.”
AI search adds another variable. It now accounts for a small percentage of hotel reservations, coming in at 4% in 2026, per the State of Distribution report. However, 55% of hotels report no change or only a minor change to distribution strategy.
Moore said hotels should take note of how they appear in AI-generated answers, "much in the same way the industry learned to monitor organic search and metasearch."
There's a CRM implication as well, Tarasova said.
"Discovery may become more automated, while the direct relationship becomes more valuable. Hotels need strong first-party data and a clear identity strategy so that when a guest does interact directly, they can recognize them and continue that relationship across channels," she said.
Looking ahead
Even though it represents a small share of bookings and has yet to really pack a punch in terms of manual work, AI is useful for solving specific problems, Moore said.
This includes content creation, data synthesizing and reporting, personalization and merchandising improvements, and customer service, among other areas.
Moving forward, hotels will want to evaluate their processes.
"The next phase will be less about asking, 'Where can we use AI?' and more about asking, 'Which workflows or customer journeys should operate differently because AI now exists?'" Moore said.
And even with hotel commercial teams working more closely together, fragmented data remains an issue that needs to be addressed to effectively leverage the new tech.
"You end up spending a lot of time integrating, matching and cleaning data before you can even start talking about personalization. And in my opinion, this is an absolute must before we can properly consider AI tools for hospitality," Tarasova said.
"I think the industry is therefore moving away from the idea that we need another platform for every problem. The priority now should be making the existing ecosystem work together and deciding which system owns which data."