Hotels have not adopted AI. Their staff have. Those are two very different things, and the gap between them is where most of the money will be wasted over the next year.
New research for Destination AI, reported by Cathy Walsh on PhocusWire, surveyed 107 hotel company leaders in August and September. Seventy-two percent said they get their AI capabilities from general tools such as ChatGPT, Gemini and Copilot. Only 33% are using AI features from their existing hotel software vendors. Just 16% are buying from AI-first startups.
Read that again. The tools hotels lean on most were not built for hotels at all.
A ChatGPT login is not a strategy
Here is the number that matters most. Sixty-three percent of respondents use AI regularly at work. Only 13% say their organisation has a clear strategy, formal governance and scaled execution across functions.
Another 30% have strategy and governance on paper, but execution is fragmented. And 26% have no formal strategy at all, just ad hoc experiments.
So the typical hotel company today looks like this: lots of people quietly using a chatbot to write emails, summarise reports and draft social posts, with very little joining it up. That is not transformation. That is a productivity habit.
Nothing wrong with habits. But a habit does not fix your rate parity, your guest data or why AI assistants describe your property wrongly. Those problems sit in systems, not in prompts.
Bottom-up is working. For now.
The survey found that success is more often driven by people than by the boardroom. Thirty-four percent pointed to an experimentation culture and 32% to bottom-up innovation. Top-down approaches were less common: a dedicated AI lead (24%) and executive mandates (14%).
I will say something slightly unpopular. That is great for getting started, and terrible for scaling.
Bottom-up experiments give you quick wins and enthusiastic teams. They also give you 40 different prompts for the same job, no shared rules on guest data, and nobody who owns the outcome.
The hotels that win from here will keep the curiosity but add an owner. Somebody has to decide which experiments become how the business works.
The vendors should be worried
If I ran a hotel tech company, this survey would keep me up at night. General tools are beating built-in AI features by more than two to one in adoption. And 41% of hotel leaders say they are building or customising their own solutions in-house.
Spending signals are mixed rather than bleak. Over the next 12 months, 27% expect to spend more with hotel software providers, 33% about the same and 17% less. That is not a collapse. But it is a warning: “we added an AI button” is not enough when the hotel already pays for ChatGPT.
Vendors still hold one huge advantage, though. They hold the data. A general chatbot does not know your inventory, your rates, your guest history or your policies unless someone connects it. The vendor that makes that data clean, open and usable by any AI will be worth more than the vendor with the shiniest assistant.
Nobody is counting
The weakest part of the picture is measurement. Thirty-five percent of respondents do not measure return on investment at all. Only 14% use defined performance indicators measured before and after deployment.
Where hotels do see results, internal gains lead. Thirty-eight percent named operational efficiency as the area with the strongest results, while 28% pointed to commercial performance.
Efficiency is the easy win. Commercial impact is the hard one, and it is where the guest is. As I wrote last week, AI referral traffic is the wrong number to watch. If you are not measuring before and after, you cannot tell whether AI is helping you sell rooms or just helping you write nicer meeting notes.
And while teams are busy using AI internally, AI outside the building is forming its own view of your hotel. If you are an independent, that view may be close to nothing, which is why AI has barely heard of many independent hotels.
What I would do this quarter
- Audit what people already use. Find out which general tools staff rely on and for what. You probably have a strategy hiding in those habits.
- Name an owner. Not to kill experiments, but to turn the good ones into standard practice.
- Set a baseline before the next rollout. Pick two or three KPIs, measure them now, and measure them again in 90 days.
- Push vendors on data access. Ask how your rates, content and policies can feed any AI tool, not just theirs.
- Point AI outward as well as inward. Check how assistants describe your property, and fix the source content they rely on.
If you want help building an AI and search plan that goes beyond chatbot logins, my travel AI and GEO services page sets out how I work with hotels, airlines and destinations.
The debate
Hotels have embraced AI as individuals. They have not yet embraced it as businesses. Seventy-two percent using general tools, 13% with a scaled strategy and 35% not measuring ROI is not an AI industry. It is an AI hobby.
So which matters more for hotels right now: freedom to experiment, or someone with the authority to say “this is how we use AI here”?
Follow the newsletter: Dan Taylor on LinkedIn