There is a familiar moment in hotel operations. The monthly report is ready, the dashboard looks clean, and the revenue meeting is about to start. Occupancy is up. Average daily rate looks healthy. Someone asks whether it is time to raise prices.
Then the finance team points out that the room revenue in the dashboard does not match the books.
It is tempting to call this a spreadsheet problem. Usually, it is more interesting than that. The people involved may all be working with perfectly reasonable numbers, but they are measuring different things.
A hotel might record what the guest paid, what the booking engine confirmed, what the property earned from room sales, and what the bank actually received. Those figures are related. They are not interchangeable.
A case study: one small hotel, two very different RevPAR figures
Consider a fictional 20-room independent hotel, which we will call Harbour House. The figures below are illustrative, not results from a real customer or a SUMS Studio deployment.
In April, Harbour House has 600 available room-nights and 420 occupied room-nights. Its reservation export contains Rp 462.000.000 in guest charges. At first glance, the spreadsheet produces an ADR of Rp 1.100.000 and RevPAR of Rp 770.000.
But the export includes Rp 42.000.000 in taxes and service charges that the property's reporting policy excludes from room revenue. Actual room revenue for these calculations is Rp 420.000.000. That changes ADR to Rp 1.000.000 and RevPAR to Rp 700.000. Occupancy stays at 70%.
A Rp 70.000 RevPAR difference may not sound dramatic. Across 600 available room-nights, however, it represents a Rp 42.000.000 difference in the revenue basis used to assess performance. More importantly, the team could be making pricing decisions from an inflated benchmark.
Now imagine the same export also contains a summary row that totals the month's reservations. If that row is added to the individual transactions, the distortion becomes much larger. No one has made a difficult mathematical error. The spreadsheet has simply been asked to add together numbers that were never meant to be added together.
Where the confusion begins
Hotel data rarely arrives in the neat format shown in a product demonstration. One property works from a property management system. Another combines booking-engine exports with an accountant's spreadsheet. A villa operator may have direct bookings in one file, OTA reservations in another, and payment records somewhere else.
Even within one file, “revenue” might mean total guest charges, room-only revenue, cash collected or revenue recognized for the stay period. The date field might refer to booking creation, arrival, departure or payment. A canceled booking may still appear beside completed stays.
These details matter because ADR, occupancy and RevPAR are not just formulas. They are agreements about what belongs in the numerator, what belongs in the denominator and which dates count.
What AI should actually do
The most useful role for AI here is not to generate a confident answer from uncertain inputs. It is to help the team find out what those inputs mean.
Imagine uploading Harbour House's export and receiving a short set of questions: “Does this amount include tax and service charges?” “This row appears to be a monthly subtotal. Should it be excluded from transaction totals?” “Three reservations have inconsistent stay dates. Can you confirm them?”
That is much more valuable than silently filling every gap. AI can suggest column mappings, identify likely duplicate totals and flag anomalies. But once the definitions are agreed, calculations should follow clear, testable rules.
A better question for the next revenue meeting
Before discussing whether ADR should rise next month, ask the team to explain where last month's ADR came from. Which revenue categories were included? Were canceled stays excluded consistently? Were room-nights counted against the right inventory? Did the export contain subtotals?
At SUMS Studio, this is the problem space that shapes our approach to hospitality business intelligence: helping hotel and villa operators move from messy operational records toward clearer, more defensible decisions.
The next generation of hospitality analytics will be judged by more than how quickly it draws a dashboard. It will be judged by whether a hotelier can trust the conversation that follows.