G3 RMS Science
Revenue Management

How It’s Made: The Science Behind G3 RMS

By , Director, Product Management & Solution Success

How does our RMS work? Pay attention to that man behind the curtain—as he opens the hood of the industry’s leading revenue science engine.

Since early 2020, the general population has gotten a crash course in the scientific process as we followed public health and vaccine development. While some remain skeptical, many of us have marveled at the rapid development of highly effective vaccines, confidently assured by the exhaustive research, painstaking testing, and peer reviewed studies that went into their creation.

Alas, I’m no public health expert, just a humble software builder and predictive analytics evangelist. But I know the difficulty in conveying the value of complex, algorithmic processes and systems and how easy it is for one to argue their analytics are the best. It may be easy to build an algorithm, but what really matters is that they are created and tested using the same rigorous scientific processes. It’s this process that supports the creation of the predictive analytics that drive IDeaS’ decision systems.

The Rigor of Revenue Science

IDeaS set out to build G3 Revenue Management System (RMS) with a vision to reinvent what’s possible for revenue technology. Other systems may do a decent job enabling data-driven, dynamic pricing, but they don’t go that extra mile. Most revenue optimization algorithms in the marketplace are limited by overly simplified assumptions about how the world works and neglect the full suite of controls needed to optimize business. Their statistical approach and data sources just aren’t robust enough, and so there will always be circumstances that require manual intervention or money left on the table.

For IDeaS to stand behind a novel solution, we challenged ourselves to build something that could account for real-world business realities without requiring hands-on attention at each property. Results had to be statistically proven, backed by research, statistical simulation and peer review from academia. Full automation eliminates the possibility for user error, creates a framework where our RMS can react as conditions change, and limits the need for human resource consumption. With a clear mission, we rolled up our sleeves and got to work.

Critical to the success of a solution is accounting for the unique nature of the hospitality business, based on decades of field research and operational understanding. We calibrated the system to incorporate comprehensive datasets and account for limited volumes of data, enabling it to create a more complete picture of unconstrained demand and market conditions, while considering a full range of key factors, such as special events, competitor pricing, days to arrival, and many more.

Many solutions use a simple approach of having one or two forecasting approaches—one for transient and one for group. IDeaS eschews this method and amps up the horsepower by leveraging over 100 models tuned to different types of hotel business. Consider the many business models hotels have, such as wholesale/crew allotments vs. typical group business, or accounting for flexible rate segments or linked rates.

IDeaS also intentionally avoids the typical sequential approach of trying to optimize room and rate availability, and then setting price, because this method makes the invalid assumption that demand does not change as prices change.

IDeaS’ proven approach folds all key data sources directly into optimization (competitor pricing, for example, is accounted for in optimization, as opposed to applying it simply as pricing rules after the RMS sets a price), optimizes all room types optimally, and avoids rules in doing so. IDeaS favors an accurate dynamic programming-based optimization, as opposed to simpler deterministic approaches that assume the demand forecast and other calibrations and assumptions are perfect.

The results of these efforts empower hoteliers to optimize pricing and forecasting outcomes like never before, but our development process didn’t end with creating the most complex, holistic, research-based algorithm we could. Our work was then subjected to multiple rounds of unbiased peer review—because that’s science for you. We turned to hospitality industry experts and, importantly, academia to validate our approach and ensure the system was well-primed to adapt to uncertainty and produce consistently beneficial results.

And after that, it was ready for release, right? Nope, after the likes of Cornell, Columbia, Georgia Tech, and others had signed off, we moved onto a rigorous testing phase. This necessitated another development project of its own. Our team created custom-built simulation tools we could use to stress-test the system and ensure it would react appropriately to real-world data, and that it would deliver results over other best-in-class revenue technology (a practice we have continued through real installations and reviewed results against many other solutions). Finally, we conducted live pilot tests in controlled, statistically matched properties to confirm outcomes would benefit our clients.

Machine + People Learning

After years of scientific scrutiny and engineering perfectionism, G3 RMS was launched—and I’m proud to say it only gets better with age. This is because it’s now been implemented at tens of thousands of properties, and we can use the data collected to constantly improve the algorithms, ensuring the forecasting approaches assigned by our system automatically benefit the property by producing the best match to the data conditions observed and that the system responds as conditions change.

So, how does our G3 RMS solution work in real time? Our team of data scientists—including more than 25 PhDs—continuously works to improve the forecasting and optimization processes, and we create tools and verify the properties that show any sign of underperformance. And on top of all these smart people and processes working to make revenue performance better for all the clients who trust us with their business, the system is always working to improve upon itself, regularly recalibrating and adjusting, without human intervention.

The groundbreaking artificial intelligence in G3 RMS allows each implementation to autocorrect, individually, as needed and continuously learn about the property at which it’s installed (and how its controls are impacting in the market it is supporting), applying and adjusting models to produce the best results. Humans needn’t be involved in deciding which models and parameters are selected or how data is incorporated. These are areas a well-designed solution will always perform best, and it’s these automated processes, combined with performance simulations and academic research (not forgetting peer review), that give our users more confidence in the system’s decisions and more time back in their day—and drive more profit for their hotels.

IDeaS’ advanced analytics allow for data to back up each level of decision-making so the system is never left to make assumptions about things that can greatly impact potential revenue opportunities (e.g., cancellations, upgrades, room-type dimensions, booking extensions, competitor impacts, length-of-stay effects, etc.).

G3 RMS is the only revenue solution that optimizes pricing for all key products and for each length of stay and creates dynamic-rate restrictions and overbooking controls to maximize revenue effectively, without setting simple business rules. This is why we have been able to demonstrate statistical revenue benefits over other solutions that can only optimize these decisions independently. We continue to drive innovation in these areas and are excited to continue to bring pricing capabilities to new types of business and business models (e.g., enhanced approaches to pricing for longer stay business, automation at scale, and supporting hotels in their path to profit maximization).

Scientific Process

Gotta Have Faith

Science can be fascinating and awe inspiring, but day-to-day progress can be hard to recognize (that said, I am always very passionate about it and always welcome your questions!). The scientific process takes time. It requires rigor and due process, and it can sometimes take time to bring algorithms that inspire confidence to bear, especially when they are applied to thousands of properties’ business decisions. And at IDeaS, we wouldn’t have it any other way. This type of process is critical in this space, and we do not take our clients’ trust for granted.

We are responsible for the performance of our solutions and, ultimately, the success of our clients. That’s why we don’t cut corners, and we certainly don’t leave anything to chance or human intuition—no offense, humans. Because of these guiding principles, an IDeaS RMS is future-proof, fully automated, and truly science-backed. And even if all of that still isn’t convincing enough for you, the real proof is in the ROI.

Director, Product Management & Solution Success

Stephen Hambleton supports hotels in driving higher revenue through revenue management technology powered by high-performance forecasting and analytics. He specializes in big-data solutions and continual technological innovations that drive revenue and profitability improvements.

Related Resources

Hungry for more?

At IDeaS, we’ve always got an ear to the ground and our fingers on the keyboard, ever-ready to share our latest learnings, data, trends, and happenings with you, dear reader.

See all blogs