5 Sep 2026

Occupancy math: how to forecast in-room screen revenue before you sign anything

Most in-room screen conversations stall at the same place: the operator asks what a screen earns, the vendor quotes a range, and nobody can check the number. The honest answer is that per-room revenue is an output of four of your own numbers, and you can estimate it yourself in ten minutes. Here is the model, the realistic ranges for each input, and the three places operators most often fool themselves.

The four inputs

Per-room monthly revenue is driven by:

1. Guest nights: nights the room is actually occupied, not nights it exists. A 30-day month at 70% occupancy is 21 guest nights.
2. Screen-on hours per guest night: how long the TV is on and showing your surface, not standby. In short-stay units this typically lands between 2 and 4 hours, higher in apartment-style stays where the TV doubles as living-room furniture.
3. Ad fill rate: the share of available ad slots that actually serve a paid ad. This is the number operators overestimate most. Household programmatic fill for niche, geo-targeted inventory commonly sits between 20% and 60%, not the 90%+ people assume.
4. Effective revenue per filled slot: what the demand actually pays for your audience, your geography and your screen size. This varies by an order of magnitude between a Gulf capital and a secondary European city.

Multiply them out and you get the ad line. The upsell line, things like late checkout sold through the same screen, behaves differently: it converts on guest intent, not screen hours, and we treat it separately because the offer menu, not ad delivery, drives it.

A worked example, with deliberately boring numbers

Take a 40-unit short-stay property, 70% occupancy, 3 screen-on hours per guest night, 35% fill, and a mid-range effective payout. The arithmetic is 21 guest nights times 3 hours times a modest number of slots per hour, then times fill and payout. Run it and the monthly ad revenue per room lands in the low single-digit dollars per screen month in weak demand markets, and meaningfully higher in strong ones. Multiply by 40 rooms and annualize, and you get the real decision figure: enough to matter as pure margin on screens you already own, and not enough to justify new hardware, which is exactly why we run on the TVs already in the rooms.

The point of the exercise is not the number itself. It is that every assumption is now visible, so when a pilot reports actuals you can see which input you got wrong, instead of arguing about the total.

The three numbers operators overestimate

Fill rate. Vendors quote campaign availability, operators hear guaranteed delivery. In a pilot, measure served impressions against available slots yourself; that ratio is your fill rate and it is the single biggest driver of the gap between forecasts and reality.

Screen-on time. Guest-night TV hours in hotels skew lower than people expect for short stays under three nights: guests arrive late, sleep, and leave. Pull actual check-in to check-out times from your PMS and discount accordingly.

Effective payout. Advertised CPMs for premium connected-TV inventory describe broad, measurable audiences. In-room inventory in a single building is niche by definition; use the low end of any quoted range for planning and treat upside as upside.

What a pilot should therefore prove

A pilot that only reports revenue proves nothing, because a good week proves nothing in any small sample. What it should return is each input measured against your property: real occupancy of the piloted rooms, metered screen-on time, actual fill rate and actual effective payout. Four to six weeks on a handful of rooms is enough to pin all four, and the results extrapolate linearly across identical rooms, which a revenue total never does. Our network readiness checklist covers what to verify before those rooms can even run the pilot cleanly.

The two revenue lines do not scale together

Operators sometimes add the two lines and divide, which hides how differently they behave. Ad revenue scales with guest nights and screen hours: double occupancy and the ad line roughly doubles. Upsell revenue scales with guest intent and offer quality: the same 21 guest nights can produce three times the upsell take if the offer menu changes, with zero change in screen hours. That asymmetry matters for forecasting honesty. Model the ad line bottom-up from the four inputs above, because physics constrains it. Model the upsell line from a pilot's measured conversion rate per guest night, because nothing else predicts it. A blended per-room number that mixes both will be wrong in both directions at once, and it will be impossible to tell later which assumption failed.

Keeping the forecast honest

Build the forecast as three columns: conservative, expected, strong. Use the conservative column for any decision that spends money, and the expected column for decisions that only spend attention. If the conservative column still clears your cost of running the screens, which for existing Smart TVs is close to zero, the decision is easy. If it only clears in the strong column, the pilot exists to tell you which column you actually live in, and that is a question worth answering before you scale anything.