16 Nov Cornell Report Highlights AI Challenges in Hotel Revenue
Hotels have bought plenty of technology over the past decade. A new Cornell report argues that AI will not fix what those systems already leave broken. The bottleneck is still fragmented data. A smarter tool layered on top of it won’t change that.
Key takeaways
- Fragmented data and systems are the biggest obstacle to AI adoption in hotel revenue management, according to the Cornell study.
- Revenue teams still measure success by RevPAR Index, not profit, which pushes decisions in the wrong direction.
- Staff override system recommendations often enough that the tools never get a fair test.
- Fixing data and incentives comes before buying more software.
Why AI keeps hitting the same wall
Sheryl Kimes and Breffni Noone, writing for the Cornell Center for Hospitality Research, gathered input from 23 senior industry leaders in interviews. They also surveyed 224 revenue management professionals around the world. Their conclusion is blunt. AI is exposing old problems in hotel revenue management, not creating new ones. The summary we read doesn’t break out the survey percentages, so treat the specifics as the authors’ findings rather than our own measurement.
The three main findings are uneven commercial integration, metrics that favor RevPAR Index over profitability, and fragmented data. RevPAR, or revenue per available room, is the basic measure of how well a hotel fills its rooms at the rates it charges. RevPAR Index compares that figure against a hotel’s competitive set, the group of nearby properties it is measured against.
Here’s the catch. If a revenue manager is scored on RevPAR Index, a tool that pushes for a slightly lower rate to win occupancy can look like the right call. That holds even when the property ends up with less profit. The software isn’t the problem. The scorecard is.
Operators at Otelier’s customer summit made a similar point. Their takeaways, reported by HospitalityNet, describe senior operators wanting to move away from dashboards. They want operational consoles that tell staff what to do next. That is a different request from more data on a screen.
What this means for your property
Most hotel teams don’t have one system. They have a property management system, an accounting package, a purchasing tool, payroll, point-of-sale terminals and a forecasting model. A person often moves numbers between them by hand. Lodging Magazine’s Omri Shalev calls this staff spending whole days shuttling data from one machine to another, work people were never meant to do.
That’s the work worth auditing first. Before you sign another contract, list where the same number gets typed twice. Look at who reconciles the forecast against the rate plan, and how long it takes.
Pay attention to overrides, too. The Cornell authors list frequent overrides of system recommendations among the barriers. When a revenue manager rejects the suggested rate every morning, the recommendation is not being tested. The team needs to know why. That answer usually lives in the scorecard or in trust, not in the algorithm.
Fix the data and the incentives first
The report’s message is that technology alone won’t deliver meaningful results. The authors argue the gains depend on organizational work. That means aligning commercial goals, clearing up who decides what, and improving the data feeding the models. Those steps are slower than a software rollout, and they are less visible to ownership.
They are also where most of the money sits. A hotel that measures profit per room, not just its share of the market, has a reason to let a tool make a better call. For more on tying systems to guest outcomes, see our piece on using guest data to improve satisfaction.
Peak works with ownership groups on exactly this kind of cleanup.
From how a property’s numbers get reported to how its operating systems hand off work, we look at the whole chain. Our hotel management approach starts with the assessment. It does not start with the software order.
A tool that can’t see the whole property will only make a faster guess.
Questions people ask
What is RevPAR Index?
RevPAR Index compares a hotel’s revenue per available room with the average of its competitive set. A score above the average means the hotel is taking a larger share of the revenue than its peers. It measures market share, which is why the Cornell authors say it can pull attention away from profit.
Why does fragmented data slow down AI in hotels?
AI tools need consistent inputs, such as forecasts, inventory counts and rate history. When those sit in separate systems that don’t share information, the tool works from partial data. The Cornell study names fragmented data as the most significant obstacle to adoption.
Should a small property buy AI tools now?
Start with the data and the reporting first, then test one tool against a clear question. The sources don’t set a cost or a timeline for a small property, so match any spend to the operating problem you can measure.
If your property’s systems don’t talk to each other, the fix starts before the software. Talk with Peak about a property assessment that maps where your data breaks down and what to change first.
See how Peak approaches hotel management and property assessments

No Comments