If you knew with 95% accuracy how full your hotel would be next month or even six months from now, what would you do differently?
At Cloudbeds, we built an AI-powered forecasting engine to help hoteliers answer exactly that. With Cloudbeds Revenue Intelligence, powered by Signals, our goal is simple: give hotels accurate, real-time forecasts so they can confidently plan for the future.
To make this possible, we partnered with Snowflake, the enterprise data platform trusted by some of the world’s most innovative companies, from tech to finance and retail. Their modern architecture gave us the speed, scalability, and data-sharing capabilities we needed to process 4 billion data signals in real time and forecast occupancy with up to 95% accuracy.
The complexity behind hotel forecasting
Forecasting in hospitality is uniquely challenging. No two properties behave the same, and demand changes constantly based on a range of factors:
- Guest booking patterns
- Local events, holidays, and weather
- Property type, size, and location
- Seasonality
- Competitive sets
To deliver accurate forecasts, your revenue management system needs to understand how all these variables interact, not just in general, but for your specific property, market, and guest behaviors.
Why most forecasting models fall short
Traditional RMS platforms rely mainly on historical data, using a small set of variables (like past occupancy rates and booking pace) to make predictions. These systems often assume demand follows the same patterns year over year and struggle to adapt to new situations.
A great example of this was the COVID-19 pandemic, where historical trends became virtually meaningless overnight. Travel restrictions, shifting behavior, and unpredictable reopenings rendered most forecasting models ineffective.
Even outside of extreme events, this limitation shows up in subtle but important ways:
- A new festival or major event throws off expected demand
- A competitor opens (or closes), changing market share dynamics
- Guest booking windows shift due to economic or political changes (ie., recent tariffs)
- Bad weather hits a beachside property, impacting demand
- Major airline cancellations impact same-day and next-day arrivals
Most systems can’t process these variables in real time, and they certainly can’t scale across thousands of properties. That’s where Snowflake and Cloudbeds change everything.
Why we partnered with Snowflake
At Cloudbeds, we knew hoteliers needed more than just a static projection. They needed a forecasting engine that could:
- Incorporate data unique to their property, market, and guest behavior
- React to real-time changes in pace, demand signals, and external events
- Surface actionable insights to improve occupancy, revenue, and team alignment
To deliver that, we at Cloudbeds required a platform that could support not just thousands of custom models, but billions of data points flowing in from dozens of sources, refreshed constantly. That’s why we chose Snowflake.
We needed a platform that could scale with our ambition. With Snowflake, we built an infrastructure that lets us prototype, train, and launch forecasting models in hours—not days. It became our single source of truth, giving every team—from finance to data science—real-time access to governed, reliable data.
Behind the scenes, Snowflake powers the data infrastructure that makes Cloudbeds Revenue Intelligence possible. It enables us to:
- Ingest and process massive, real-time data streams
- Train and deploy thousands of models in parallel
- Support experimentation across segments, markets, and forecast windows
- Store, version, and reuse powerful forecasting features
- Visualize performance across properties with dynamic dashboards
And because Snowflake is built with enterprise-grade security and compliance at its core, our customers can trust that their data is fully protected, governed, encrypted, and handled according to the highest industry standards.
What makes our approach different
So, how does Cloudbeds Revenue Intelligence show up for hotels compared to other systems?
1. Always up to date
Traditional forecasting systems are built on batch data, updated once a day, or even once a week. By the time you see the numbers, the demand picture may have already changed.
Revenue Intelligence uses Snowflake to stream billions of data points every hour, pulling in real-time booking trends, local events, weather, holidays, and more.
What this means: Your forecast reflects what’s happening right now, not what was true yesterday.
2. Personalized to your property
Most RMS platforms apply the same forecasting rules to every hotel. These generic models miss the nuances that make your business unique.
Cloudbeds builds custom models based on your property’s booking curve, guest behavior, location, and seasonality—then continuously refines them with new data.
What this means: You get forecasts tailored to your property, not a generic industry average.
3. Fast, scalable, and responsive
Revenue management systems can take hours to process and update forecasts, especially across multiple properties.
With Snowflake’s cloud-native infrastructure, we’re able to train and deploy thousands of models in parallel, allowing our team to rapidly iterate and continuously improve forecasting accuracy across property types and markets.
For day-to-day forecasting, Snowflake’s partitioned inference capabilities allow us to generate updated forecasts in seconds, delivering real-time insights to hotels.
What this means: You can react faster, test new strategies more often, and stay ahead of demand shifts in real time.
4. Built to improve over time
Many traditional models stay static until someone manually intervenes. They don’t get smarter as conditions change.
Revenue Intelligence continuously evaluates forecast performance by lead time, day of week, property type, and region, learning and adapting automatically.
What this means: Your forecasts get more accurate the longer you use the system.
What this looks like in practice
Let’s say you’re forecasting four months out. Here’s how it might look.
120 days out: You’re pacing at 15% occupancy—slightly behind pace for this season.
→ You adjust your metasearch bids to boost visibility on high-converting channels, refresh your paid search and social ads with seasonal messaging, and launch an email campaign targeting past guests who traveled during the same period last year.
90 days out: A major event is announced nearby, and similar properties see a surge in demand.
→ You increase rates by 8%, update availability across channels, and shift ad messaging to appeal to event attendees.
60 days out: A key competitor raises their rates by 10%.
→ You hold your current rate to capture additional bookings, or increase slightly while bundling added value (e.g., free breakfast or late checkout).
30 days out: A rainy weather forecast is predicted, softening demand.
→ You re-target local and drive-market guests with flexible stay offers and indoor amenity promotions.
Final forecast (week of): System projects 87% occupancy with strong short-window demand.
→ You restrict discounts, optimize remaining inventory with targeted upsells, and prepare operations for full capacity.
With Cloudbeds, forecasting becomes a proactive, day-to-day decision-making tool, not just a report. It helps your team react faster, price smarter, and maximize every window of opportunity.
What better forecasting means for your hotel
With Cloudbeds Revenue Intelligence, you’re not just forecasting, you’re unlocking a smarter way to run your business.
For revenue managers, that means having real-time, forward-looking data at your fingertips to make faster, more confident decisions. You can shift from reactive pricing to proactive strategy, quickly identify pickup trends, and adjust rates or campaigns before your compset even notices the change.
For hotel operators and owners, it means higher occupancy, better rate integrity, and improved profitability across the board. With forecasts that adapt daily, your team can better align staffing, marketing, and inventory decisions to actual demand, not outdated reports or gut feel.
This looks like:
- Optimized pricing strategies
- Faster response to market shifts
- Outperformance of your compset
- Better team alignment across revenue and marketing
- More accurate staffing
- Improved budget planning
- Scalability across properties
Why Snowflake is our forecasting foundation
Achieving 95% forecasting accuracy across a global portfolio doesn’t happen by accident. It takes more than AI—it takes infrastructure designed to handle complexity at scale.
That’s why Snowflake is at the core. Its real-time processing power, flexible data sharing, and scalable architecture allow us to move faster, experiment more, and deliver insights other systems can’t match.
By consolidating all our data and ML workflows onto Snowflake, we’ve set a new standard in the hospitality industry for model accuracy. Snowflake’s scalability significantly slashed our training times, while its efficient partitioned inference has led to substantial cost savings in compute. It laid the foundation for Signals.
Together, Cloudbeds and Snowflake are redefining what forecasting can do for hospitality, giving hotels not just predictions, but powerful, profitable decisions.
