Insights · Data

Why your hotel numbers don't match — and why nobody is lying

By the YMME team·8 min read·September 2026

Ask four people in a hotel group what last month’s average rate was. You will get four answers, and every one of them can be defended. The property system counts one way. Revenue management excludes a segment. Finance uses the definition the auditors signed. The board sees whichever number made it into the deck. Nobody is lying. The group has never decided, once, what the number means.

Quick answer. When two reports disagree, the cause is almost always one of four unwritten decisions: which system owns the figure, how the figure is defined, where the cut-off sits (business date or calendar date, gross or net), and how it is rounded. None of these is a calculation error, so checking the arithmetic does not help. Write each decision down once, in the model, and make every screen read from it.

A mismatch is a decision, not a mistake

The instinct when two numbers disagree is to find the wrong one. Twenty minutes into the meeting, both turn out to be right. The revenue manager’s rate excludes complimentary rooms; the controller’s includes them. The PMS reports by business date; the accounting export reports by calendar date. One card rounds to a whole per cent; the table beside it shows a decimal.

Each of these is a legitimate choice. The trouble is that each was made separately, by whoever built that report, and recorded nowhere the next person could find it. So the next report makes the choice again, differently, and the disagreement is reborn. The fix is not a better spreadsheet. It is to make each choice once and give it a home.

Which system owns the figure

Every figure has one source that is authoritative for it, and the reporting layer has to know which. Money comes from the ledger. Occupancy and rate come from the night audit. Covers come from the till. Reading occupancy out of financial postings sounds efficient and gives a wrong answer: complimentary and house rooms are occupied but post no revenue, so the hotel looks emptier than it was, every month, by a margin nobody can explain.

The rule is easy to state and hard to keep: every figure comes from the system that owns it, even when another system has a number that looks the same. Where two channels reach the same vendor, or two properties run different versions of the same PMS, the model records which one wins. For a property that changes system, it records the date from which the new one takes over.

What the figure means

Average rate: with or without complimentary rooms? RevPAR: on rooms available, or on rooms available less those out of order? Food revenue: with the breakfast that is bundled into the room rate, or without? Each has a defensible answer. The group needs one of them, written down, and every screen that shows the measure has to read from that one definition.

This is where USALI does its work. The Uniform System of Accounts for the Lodging Industry is the layout that lenders, auditors and buyers expect, and its twelfth revised edition took effect on 1 January 2026. Reporting to it does not settle every operational definition, but it settles the ones a bank will argue about, and it makes one hotel comparable with another and with itself a year ago.

Where the cut-off sits

Business date or calendar date is the classic. A hotel’s day ends at night audit, not at midnight, so a Saturday in the PMS is not the same Saturday as in the bank statement. Gross or net is the other: with or without service charge, city tax, the agent’s commission. Two reports that draw the line differently will never reconcile, and the difference will look random because it is not constant.

The discipline is one cut-off per screen. A screen reports either by business date or by calendar date, either gross or net, and says so in its heading. Mixing two on one screen because the data happened to be convenient is how people learn to stop reading it.

Owner. Which system is authoritative for this figure, and which takes over after a migration.
Definition. What is in and what is out: complimentary rooms, out-of-order rooms, bundled breakfast, service charge.
Cut-off. Business date or calendar date; gross or net; one choice per screen, named in the heading.
Precision. How many decimals the figure carries, and the same number of them everywhere it appears.

How it is rounded

Rounding is the one that eats the most meeting time. A card reads 78% of budget; the control table beside it reads 77.7%. Both are correct. The meeting spends twenty minutes on which one is wrong. So the same measure is rounded the same way on every surface it appears on, and the rule sits in the model, not in each chart.

Currency follows the same logic. A property that earns in one currency and pays its suppliers in another does not have a total; it has two figures, and a reporting layer that adds them at a rate somebody typed into a cell has produced a number no auditor will accept.

Two correct numbers that disagree cost more than one wrong number. The wrong one gets fixed. The two correct ones get argued about every month.

Comparing on a base that has moved

One more source of mismatch lives in time. A group reclassifies half board out of food and beverage and into the room rate. Accommodation revenue rises by a third without a single extra guest. Reported unadjusted, that is a triumph and a bonus. Reported properly, it is a reclassification, and the prior year has to be restated on the new basis before the two are placed side by side.

A reporting layer either knows about reclassifications or it does not. If it does not, every year-on-year comparison after a change in the chart of accounts is comparing two different bases and calling the difference growth.

How this is built in Score. In YMME Score each measure is defined once, in the model — owner, definition, cut-off, precision — and every screen, card, table and emailed summary reads from that definition. Where a group has changed its chart of accounts, comparable series are rebuilt before the comparison rather than explained afterwards. The screens are illustrative; the four decisions are the product.