HotelRMS — AI-native hotel revenue management
HotelRMS
Autopilot Benchmark

How often do hotel revenue managers
accept HotelRMS recommendations?

We publish a rolling 90-day, network-wide measure of how often operators apply our price recommendations without manual override. Methodology and sample-size threshold are disclosed below. The live number is shared during a demo conversation so we can put it in context.

Live number — by request
See where we are on the autopilot benchmark today.

We share the headline number during a short demo so it lands with context — what good looks like in revenue management, how it varies across customer types, and what your property might expect. Methodology is fully public below.

Request a demo →

No signup. Credentials emailed within one business day.

Methodology

The benchmark is intentionally simple to keep it auditable. Definitions, thresholds, and lookback window are fixed and published — we will not retro-fit them to make the number look better.

Net-autopilot score (per customer)
applied_without_override ÷ decided × 100

Where decided = applied + dismissed + expired (i.e., anything not pending). A decision is "clean" when the operator accepted the recommendation and did not adjust it before applying.

Network aggregate
Equal-weighted mean across all customers whose individual sample meets the threshold. Hyatt's score does not drown Bob's Inn's — every qualifying customer counts once.
Lookback window
Trailing 90 days. Long enough to be statistically meaningful, short enough to reflect current operator practice.
Sample threshold
Each customer must have at least 25 decided recommendations in the lookback window to be included. Customers below threshold are excluded so a handful of decisions cannot skew the published number.
What is excluded
Pending recommendations (operator has not yet decided), decisions outside the 90-day window, and any customer whose individual sample is under threshold.
Refresh + auditability
Computed server-side and cached for one hour. The methodology version is exposed in the public API response so external observers can detect any change to the formula.

Why publish this at all? Most revenue-management vendors do not. We do because the net-autopilot score is the single most honest measure of whether a system is actually earning operator trust — independent of marketing claims or case studies. A low number means our recommendations need work; a high number means revenue managers see the system as a colleague rather than a tool to fight against.