Build the Infrastructure for Data-Driven Decisions

A shortage of data is rarely the real problem in a hotel. The PMS closes every night, the channel manager shows sales channel by channel, accounting closes the books at month end. The problem is that these three sources do not speak to one another, and none of them answers the question “what should I do this week?”

A hotel whose reporting has stalled shows a few familiar symptoms:

• The same metric returns two different numbers in two reports; half the meeting goes on arguing which one is right.
• The report arrives after the month has closed — the date on which anything could have been done has long passed.
• Every department keeps its own spreadsheet; nobody sees the total.
• The report exists but nobody reads it, because there is no threshold inside it that triggers a decision.
• Budget and forecast sit in the same file; where the variance came from is invisible.
• Channels are ranked on gross revenue, and nobody notices the ranking changes once commission comes out.

The purpose of reporting is not to record the past but to decide what happens over the next 30 to 90 days. A table that explains the past belongs to accounting; a revenue management report looks forward.

Hotel reports bloat because everything measurable gets measured. In the structure I build, every metric has one question and one rhythm; a metric without a question comes out of the report.

Three of these carry the daily report on their own, and their arithmetic is simple; what creates debate at setup is never the formula but what gets put inside it.

The two formulas for RevPAR, which rooms belong in the denominator and the four mistakes made most often are set out step by step in a separate article — you can reach it from the link further down this page.

One report cannot serve everyone. Put the daily needs of operations and the owner's needs in the same table and neither gets read. The structure I build has three layers, and all three are fed by one set of definitions.

Daily — revenue and reservations

Yesterday's sales, pickup, occupancy for today and tomorrow, remaining inventory, competitor rate movements. The decision it triggers: rate and restriction updates. This report is short; it reads on one screen and needs no meeting.

Weekly — the management team

The occupancy and ADR curve for the next 8–12 weeks, pace against last year, channel share, group blocks on hold. The decision it triggers: opening or closing a promotion, shifting channel weight, pricing a group enquiry. This is the real revenue management meeting.

Monthly — management and ownership

Budget, forecast and actual side by side, GOPPAR, segment mix, net channel contribution, forecast variance. The decision it triggers: budget revision, investment and staffing plans. This should be the only report anyone waits for at month end — the other two will already have made the decisions.

All three layers must draw on one set of definitions. If “available rooms” means one thing in the daily report and another in the monthly, the meeting is spent arguing about the variance and trust in the report is gone.

1. 01Inventory of data sources — PMS, channel manager, booking engine, accounting and an RMS where one exists. One source of truth is chosen per number; the same metric arriving from two sources is what destroys confidence in a report.
2. 02A definition glossary — what available rooms covers, whether ADR is gross or net, the moment a cancellation counts, which taxes are excluded. This is the most skipped step and the one that causes the most argument; no dashboard is built before it is written down.
3. 03Automated flow — while data is copied by hand the report is both late and error-prone. Export or connection options are set up at this step.
4. 04A decision screen — a chart is there to answer a question, not to look good. Under every visual there should be an answer to “what does this trigger?”
5. 05Thresholds and alerts — rules such as “pickup negative over the last seven days” or “occupancy five points below last year for the next 14 days”. The report stops being something you look at and becomes something that warns you.
6. 06Ownership and rhythm — who produces the report, who reads it, which meeting it belongs to. A dashboard with no owner dies within two months.
7. 07Training and handover — the structure stays inside the hotel. The aim is not a report dependent on a consultant but an order the team can sustain.

1. One word, two meanings

Available room count, gross versus net ADR, the moment a cancellation is counted. When these differ between departments, the reports differ too. The problem is not technical but lexical, and the fix is to write the glossary.

2. Data copied by hand

A number pasted from one spreadsheet into another every morning makes the report dependent on one person and produces silent errors. When that person takes leave, the report stops. A metric that never gets automated does not survive in a report.

3. Budget and forecast blurred together

The budget is the promise made at the start of the year; the forecast is the most realistic view from today. Held in the same cell, the source of a variance becomes invisible: was the target optimistic, did demand fall, was the rate positioned wrongly — none of it can be answered until the two are separated.

4. The dashboard nobody reads

This is the most common outcome: the report is built, watched for a few weeks, then forgotten. The reason is almost always the same — there is no threshold inside it that triggers a decision. The screen looks good but never tells anyone to do something now.

The test of a well-built reporting structure is simple: the weekly meeting opens with the decision to be made, not with a debate about whether the numbers are right.

• A definition glossary — what each metric means and which source it comes from, in one document.
• Three reporting layers, daily, weekly and monthly — who reads each, which decision it triggers, which meeting it belongs to.
• A management dashboard — fed by your existing systems, with no manual copying.
• A threshold and alert list — which variance triggers which action.
• Channel and segment analysis — ranked on net contribution, not gross.
• A forecast-versus-budget template — a structure that separates the source of a variance.
• Handover — training for whoever produces the report, and the flow written down.

The goal is not to sell new software but to make existing data readable. The structure stays with you; if the reporting ends when the consultancy does, it was not built properly.