THERE IS A SCENE PLAYING OUT Pin hotel revenue management offices across the country. A sharp, tech-curious RM discovers that ChatGPT can draft a rate recommendation narrative in seconds, build a quick pick-up analysis from a pasted spreadsheet or generate a competitor summary from data she copied in manually.
Within a few weeks, she has built herself a small suite of personal AI shortcuts. She is faster. She is more confident. She is, by most measures, more productive.
Good news, right?
Yes and no. What she has built is real and valuable. For her. The moment she leaves, calls in sick, or simply closes that browser tab, the efficiency disappears. Her colleague in the next cluster has no idea the shortcut exists. The director of revenue management has no visibility into how it works or whether the data feeding it is accurate and the organization isn’t more productive than it was before she started experimenting. Even worse, any insights she’s gained from her improved access to information live with her alone!
This is the AI productivity trap: investing significant energy in tools that scale to exactly one person.
The Evolution Analogy and Where It Breaks Down
Every major technology shift in revenue management has required adaptation. The move from manual forecasting to spreadsheets and from spreadsheets to revenue management systems has asked professionals to learn new tools without abandoning their core expertise. A revenue manager who resisted Excel in 1995 was a revenue manager with a shrinking career. In that sense, today’s AI moment is no different – professionals who refuse to engage with these tools will be left behind.
But here is the critical distinction the analogy misses: when the industry moved to Excel and then to RMS platforms, the organization ultimately provided the tool. Microsoft built Excel. Your company purchased it, deployed it, trained on it and expected everyone to use it correctly and consistently. Perhaps when we first deployed Excel, each revenue manager built their own spreadsheets with their unique formats and styles. But we quickly realized this resulted in incorrect formals and inconsistent data, not to mention the wasted effort of many revenue managers who created their own versions of the same tool. We stopped asking the revenue manager to write the formulas from scratch or, for that matter, to build the spreadsheet application itself.
Today’s AI environment is fundamentally different. People without strong technical or analytical backgrounds often struggled to use Excel without training, and building efficient formulas required both “coding” skills and creativity. Today, the tools are readily available and the barriers to building something useful are remarkably low. You just need to describe it, and you don’t have to translate it into rows, columns and formulas. That accessibility is genuinely exciting. But it has also created a situation where organizations are quietly outsourcing their AI strategy to whoever happens to be the most technically curious person on the team, often without realizing it.
The Data Problem No One Is Talking About
Set aside the organizational scaling issues for a moment. There is a more immediate concern – when individual revenue managers build their own AI-assisted workflows, how confident are you that the underlying data is correct?
Revenue management data is notoriously complex. Service fees may or may not be included in room revenue depending on which system or table you pull from. Averages get averaged again. Comp set definitions shift. Year-over-year comparisons break when the calendar does not align. These are not edge cases. They are the daily texture of the discipline and experienced revenue managers navigate them through hard-won institutional knowledge.
When that same revenue manager prompts an AI tool to analyze data she has pasted in, the AI has no awareness of those nuances. It doesn’t know the occupancy figure in that extract already nets out the complimentary rooms. It can’t flag that she is averaging averages across segments with different weights. It will produce a confident, well-formatted answer, regardless of the input. And the revenue manager, working fast, trusting the tool may not catch the error before it informs a strategy.
There is also a data security dimension that organizations are largely ignoring. How many team members are copying proprietary performance data, forward-looking pickup reports or confidential rate strategies into free consumer AI tools? If you haven’t established a policy, the answer is almost certainly more than you think.
The Operating Model Has to Change
The deeper issue here is structural. Revenue management has long operated on a generalist model. The RM as a jack of all trades, expected to be analyst, communicator, strategist and ad hoc technologist.
That model made sense when the tools, while rudimentary, were relatively stable and the asks of them were bounded. It is straining badly in an environment where the technology is advancing faster than any individual can track.
We don’t ask revenue managers to develop their own revenue management systems. The idea would be absurd. Revenue management systems exist because the problem is complex enough to require specialized engineering, rigorous data governance and ongoing maintenance by people who do that work full time. The expectation that revenue managers will build production-grade AI solutions in their spare time, on nights and weekends, between strategy calls and owner presentations is, when you state it plainly, equally absurd.
What organizations actually need is an expertise model: a deliberate structure in which AI and technology capability sits alongside revenue strategy, analytics and commercial leadership as a distinct pillar, not a hobby project for whoever has bandwidth. That means having people, whether internal or external, who understand both the business and the technology and who work alongside revenue management teams in a purposeful, sustained way.
This isn’t a small ask. But the alternative, a patchwork of individual workarounds that no two people use the same way, built on unchecked data assumptions, invisible to leadership and impossible to maintain, is not a strategy. It is a liability dressed up as innovation.
What Production-Grade Actually Means
When we talk about AI solutions that scale, we mean something specific. A production-grade tool is built on clean, governed data. Data that has been validated, contextualized and connected to your actual systems of record. It is designed with the full organization’s workflow in mind, not one person’s preferences. It includes guardrails that prevent the most common errors: the averaged averages, the mismatched revenue definitions the comparisons that break across system boundaries. It is maintained and updated as the business changes. And it is accessible to everyone who should be using it, with appropriate training.
Building that kind of tool requires people who understand the revenue management business deeply enough to know what questions to ask, and who understand AI and data architecture deeply enough to know how to answer them reliably. Those two skill sets rarely coexist in a single person, which is precisely why the best implementations pair domain experts with technical specialists in an ongoing, collaborative relationship.
The other hidden cost of the individual productivity approach is attention. Revenue managers who spend meaningful portions of their week building and refining personal AI tools are revenue managers who are spending less time on revenue management. The goal was never to create a new class of citizen developers inside the commercial function. It was to make the commercial function better at what it does.
The Path Forward
None of this means that individual AI fluency is unimportant. Revenue managers should absolutely be comfortable with AI tools, should understand their capabilities and limitations and should be empowered to use them for appropriate tasks – drafting communications, exploring ideas and accelerating research. The medium has changed, as it always does and professionals who don’t adapt will find themselves at a disadvantage.
But individual fluency is the floor, not the ceiling. The organizations that will genuinely benefit from AI in revenue management are those that treat it as a commercial capability requiring deliberate investment, not a productivity hack to be discovered by the curious and ignored by everyone else.
That means building or accessing the right expertise. It means establishing data governance before you build tools on top of that data. It means designing solutions that work for the whole team, not just the person who built them. And it’s recognizing the goal isn’t to make a few individuals more productive. It’s to make the entire organization more capable.
Your revenue managers aren’t supposed to be coders. That isn’t why you hired them. But with the right support structure, the right expertise and the right investment, they can be the beneficiaries of tools that make their genuine skills, commercial judgment, strategic thinking and market intuition more powerful than ever.
The question is not whether AI will transform revenue management. It already has. The question is whether your organization will be the one doing the transforming, or simply watching a few of your best people do it for themselves.
