Cloudbeds Just Shipped the Openness Its Own CIOs Said Vendors Don't Deliver

Cloudbeds launched Insights, a native reporting and data-intelligence product paired with a Data API that lets hotel groups pipe operational data — RevPAR, ADR, occupancy, booking windows, payments, channel performance — straight into external BI tools and data warehouses, without waiting on scheduled refreshes or switching systems to check a number. Co-founder and COO Richard Castle framed the pitch around speed: “hotel groups run on decisions made in the moment,” and Insights pairs that live data with integrated workflows that let a team act on a report — touch a reservation, update a guest record — without leaving the platform.

The launch lands on a fault line exposed at a closed-door PMS roundtable earlier this month. Hotel CIOs there said only 4 of 7 vendor systems on the table were genuinely open, while the vendors claimed 4 of 5 offered cost-free integration — a real gap between how open PMS vendors say their platforms are and how open customers experience them. The same roundtable’s closest thing to consensus on data ownership was “whoever signs the contract owns the data.” A Data API that exports live metrics to a customer’s own warehouse is a concrete answer to that complaint — assuming it holds up in practice, which is exactly the part CIOs said they can’t verify from a launch page.

It’s also worth reading against Terence Ronson’s four-layer framing of where AI value accrues in hospitality tech: system of record, application/semantic layer, frontier model, compute. Ronson’s warning is that lock-in is migrating up from the PMS to whichever layer owns the unified, real-time data model — and Cloudbeds is building Insights and its Signals AI layer on top of “a single unified data model,” the same move Ronson says determines who “owns the guest” going forward. An export API is a real olive branch on openness. It doesn’t change who owns the semantic layer underneath it.

For operators evaluating Insights: test the Data API against a real export before taking “real-time” and “your data” at face value.