24 Units, 10 Months, One Table: service at a hotel lobby bar

What Is in This Note

This note looks at tech for 24 units over 10 months at a hotel lobby bar, written from the buyer's side of the table. The figures are modelled rather than measured: cup volumes, price per cup and cost lines follow stated assumptions, and the arithmetic is left visible so a reader can substitute their own numbers.

The Inputs, Line by Line

Cup volume is stated per unit per day. Revenue is cups x 30 days x units x price per cup, taken at 8.2. Cost is consumables, the site share, power and parts. Payback is equipment spend divided by steady monthly net. No figure here is a quotation.

The model keeps the ramp separate from the steady state: the early months carry a lower cup count, and the steady figure is the plateau the site settles into once regulars appear. A reader who expects a different ramp can replace it without disturbing anything else on the page.

Every Line, Every Month

The columns are shown separately because the argument is rarely about the total. It is about which line moves fastest when a site underperforms, and which lines were fixed before the first cup was sold.

Where the Rows Disagree

The best month in this model nets about 73621, the weakest about 14593. The gap is the site's own variability rather than a change in cost, and that gap is what a trial period exists to measure.

At the modelled plateau, payback lands at 6.2 months. Read it as a reading of these assumptions, not a forecast: halve the cup volume and the same placement reads very differently.

Consumables scale with sales, so they rarely surprise anyone. The site share and the route hours are the two lines that do surprise operators, because both are agreed before the first cup and neither moves when the site underperforms.

Reading the Rows for service

What telemetry does not report is the last metre: a drip tray gone sticky, cups stacked out of reach, a machine unplugged to charge a floor polisher.

Most service calls are not failures. They are stock, cups, a socket or a screen waiting for someone to press a button, and each one costs a visit if nobody sorts the categories.

A machine that reports its own faults changes the economics of a route: fewer surprise visits, more planned ones, and better stocking of the part that actually breaks.

The Follow-Up That Matters More

Space is the constraint nobody prices correctly. A cabinet on a two-square-metre footprint can be worth a corridor and worthless in a lobby, and the difference is rarely about the machine.

Flow beats footfall. A thousand people walking past a cabinet convert worse than two hundred who queue beside it.

The List to Take to the Site

1. Check what the dashboard reports and what it omits
2. Record the first three faults with date, part and cause
3. Decide the escalation path before the first failure, not after
4. Time how long people stand idle within three metres of the spot
5. Measure the delivery path: door widths, turns, lift depth
6. Confirm the trial period and the move-out term before delivery
7. Write the modelled cup volume into the plan and review it after month one

The Method, Applied to Your Numbers

The arithmetic is short enough to redo in an afternoon. Do it in this order:

1. Measure a week of cups in the venue, then discount the first two days as novelty traffic
2. Use the real consumable cost per cup, not the one in the brochure
3. List site share, power and parts separately so nothing hides inside a single number
4. Turn the payback into months, then into a band, and note which input the band depends on
5. Run the whole thing again at seventy percent of the volume and keep both outputs

The Venue Changes the Curve

The same cabinet produces different curves in different venues. These four patterns are the ones operators describe most often, and each changes which line of the ledger moves first.

The point of the list is not to rank venues but to predict which line will move. Match the cadence to the rhythm and the rest of the ledger behaves.

The Three Things Worth Keeping

The first thing to keep is the habit of arguing with numbers rather than with adjectives. Every payback claim decomposes into five inputs, and four of them are checkable before signing.

Second, the trial period is the cheapest instrument in the whole purchase. It converts an opinion about a site into a reading, and it costs nothing beyond a written move-out term.

Write the volume down before delivery. The common failure is not a bad margin on good volume; it is a good margin on volume that never arrives.

Where the Model Bends

The result is fragile in three places: volume below the floor, a host share above the band, and a route that cannot reach the site inside working hours. All three are checkable on a site visit.

When somebody quotes a faster payback, ask which of the five inputs differs. A shorter number is usually a different assumption, not a better machine.

Common questions

What should a dashboard actually report?

Stock depth per line, faults with timestamps, sales by hour and the machine's online state. Anything beyond that is useful only if someone reads it weekly.

Does remote monitoring reduce service visits?

It changes them rather than removing them: fewer emergency visits, more planned ones, and a better guess at which part to carry.

What do the modelled numbers actually mean?

They are the arithmetic of stated assumptions, not measurements. Every input is listed in the assumptions section so a buyer can replace it with their own.

Why show six months instead of a full year?

Because the first six months contain the decision: the ramp, the plateau, and the first sign of which cost line moves when the site underperforms.

Related reading

• Month One Is Not the Business: 13 Cups, Then 32
• Service at an exhibition centre, Written Down Instead of Guessed
• 23% of the Till Went to the Host: an exhibition centre Ledger
• Payback in 4 Months: 25 Units at a convenience store frontage, on Paper