What a hotel COO actually needs on one screen

Picture a multi-unit operator at 7:50 in the morning, coffee in hand, ten properties to cover before a 9 o’clock call with ownership. The hotel KPI dashboard loads. Forty widgets. Occupancy by day of week, a RevPAR index trend, a heat map of channel mix, a sparkline of social media sentiment, a gauge that has been stuck at 86 percent since the integration broke in March. Somewhere in there is the number that matters this morning. The operator does not have time to find it. So the dashboard gets closed, and the morning gets run off three text threads and a gut feeling instead.

This is the most expensive dashboard in the company, and nobody reads it. We have seen the pattern at portfolios of every shape. The build was not wrong. The instinct was wrong. Somebody confused “show everything we can measure” with “show what the operator decides on.” Those are different documents. A COO does not need a museum of metrics. A COO needs an operating view, and an operating view is short on purpose.

This is a teardown of that screen. What a good one-screen hotel KPI dashboard contains, what a bad one is bloated with, and the handful of numbers a multi-unit operator must see before the day starts.

What the bad version looks like

The bloated dashboard has a tell, and it is not the number of tiles. It is who it was built for. The bad version was built to impress the person who approved the budget, not the person running the floor. It answers every question anyone has ever asked, which means it answers no question the operator has this morning.

You can spot it in three ways. First, it mixes altitudes. A whole-portfolio number sits next to a single-outlet metric next to a channel-level detail, and the eye has nowhere to land. Second, it is all rear-view. Every tile reports what already happened, and not one tells the operator what is about to. Third, it has no exceptions. Forty numbers, all weighted equally, none flagging the one property that needs a decision today. A screen where everything is important is a screen where nothing is.

The good version inverts all three. It picks one altitude, the operator’s. It leads with what is going to happen, not only what did. And it surfaces the exception before the operator has to go hunting for it. Below is the anatomy.

1. The whole-building operating number

Start with one number that describes the entire operation, not one department. Most hotel dashboards open with RevPAR or occupancy because those are the numbers the rooms team lives in. But a COO does not own rooms. A COO owns the whole P&L. The first thing on the screen should be the figure that rolls the building up, total revenue against budget and against last year, with flow-through implied, not RevPAR sitting alone where it can be up while the building loses money.

This is the number the operator reads out loud on the ownership call. It has to be defined once, agreed by the controller and the revenue team, and identical to what lands in the month-end pack. If the operating number on the screen and the number in the board deck disagree, the whole dashboard is dead by the second meeting. That is the failure that kills most hospitality dashboards inside ninety days, and it starts at the top tile.

2. Flow-through to GOP

The second number is the one that separates operators from spectators. Revenue up is not the job. Revenue up that reaches the bottom is the job. Flow-through, sometimes called conversion, is how much of an incremental revenue dollar makes it to gross operating profit. A property can grow the top line all year and flatten GOP because labor and cost of sales ate every gain. The revenue tile will look great. The owner will be furious.

So the operating view carries flow-through next to revenue, period to date and trended, at the portfolio level and flagged where a property is leaking. This is the number that tells a COO where the building is working hard and getting nowhere. We have written separately about the trap of RevPAR climbing while GOP stays flat, and the reason operators get surprised by it is almost always that the dashboard showed them revenue and hid the conversion. On the one-screen view, the two numbers live side by side, always, so the question “did it reach the bottom” answers itself.

3. The pace and demand signal

Everything above is rear-view. The third item has to look forward, or the operator is driving by the mirror. A COO needs a demand signal that says where the next sixty to ninety days are heading, not just where last week landed. For rooms that is pace, on-the-books revenue for forward periods against the same point last year and against budget. For the larger full-service operation it is rooms pace plus a catering and banquet pace line, because that is the lumpy forward number nobody paces until it is too late.

The operating view does not need the whole pace curve on this screen. It needs the signal. Is the next quarter pacing ahead, behind, or flat, and is the trend improving or eroding week over week. One line, color coded, that tells the operator whether to push price, chase business, or hold. The detail lives one click down. The signal lives on the screen. A dashboard that shows a COO only what already happened is a dashboard that lets the operator find out about a soft quarter the month it arrives, instead of the quarter before.

4. Labor as a percentage of revenue

Labor is the lever the operator can actually pull this week, which is exactly why it belongs on the daily screen and most of the others do not. The number is labor cost as a percentage of revenue, by property, against standard and against the forecast the schedule was built on. Not headcount. Not hours. The percentage, because that is what flows to GOP and that is what an operator can correct inside a pay period.

The reason this earns a permanent tile is that labor moves faster than anything else on the P&L and it moves quietly. A demand forecast slips, the schedule was built three weeks ago against the old forecast, and labor as a percentage drifts up at four properties at once while every revenue tile still looks fine. The operator who sees the percentage every morning catches it on Tuesday. The operator who waits for the month-end report catches it after four weeks of bleed. The schedule should be built off a forecast that survives the week it was made, a discipline we cover in labor models that survive contact with the floor. On the operating screen, labor as a percent of revenue is the single tile that turns a number into an action this week.

5. The exceptions that need a decision today

The last item is the one most hotel KPI dashboards skip entirely, and it is the one a COO would keep if forced to delete all the others. Not a metric. A short list. The two or three things across the whole portfolio that need a decision today, surfaced by the system instead of found by the operator.

A property where flow-through dropped below threshold this week. A market segment that fell off pace hard enough to change the forecast. An outlet where labor blew through standard three days running. The exception list is the dashboard doing the operator’s scanning for it. The other four items tell a COO how the building is doing. The exception list tells a COO what to do before the 9 o’clock call. A one-screen operating view without an exception list is a status report. With one, it is a tool.

What to leave off

Everything else. Channel mix belongs to the revenue manager, not the COO’s morning. Outlet-level menu performance belongs to the F&B director and the menu engineering work, one altitude down. Social sentiment, web traffic, the comp set index that is probably lying anyway, all of it is real and none of it belongs on the screen a multi-unit operator reads before the day starts. The test is simple. If the operator cannot make a decision off it by 8am, it lives one level down, off the front page. Five things the operator decides on, defined hard and owned by name, beat forty things the operator scrolls past.

This is the operator-intelligence layer of the larger model, and it only works if the foundation is real. Five clean numbers on one screen are impossible when the building is still running on five systems that disagree. The operating number has to reconcile to the controller’s pack. Pace has to join the PMS, the RMS, and the sales-and-catering system. Labor has to tie to the same revenue the flow-through tile uses. The reason most one-screen dashboards are actually forty-widget dashboards is that nobody did the join underneath, so the build compensated for missing trust by adding more tiles. More tiles is the symptom. A connected source of truth is the cure.

That is the Operator Intelligence model. One connected source of truth underneath, an intelligence layer that reads the way an operator already thinks, and an operating cadence that makes the screen a daily habit instead of a quarterly demo. The one-screen view is the part of it the operator touches every morning. It is also the part that exposes whether the rest of it is real.

You can build this screen yourself from the five items above, and some operators should. Pick the operating number, agree it with the controller, add flow-through, the pace signal, labor as a percent, and an exception list, and delete the other thirty-five tiles without ceremony. If you would rather it just existed, tied to clean numbers, current every morning without someone rebuilding it in a spreadsheet on Sundays, that is the kind of thing we build. Either way, the screen a COO actually needs is shorter than the one most companies paid for. That is the whole point.