What one more guest is actually worth: 27 pesos, not 100

The number most owners use is 3.7 times too big

Ask an owner in Oaxaca whether a discounted bed is still worth selling and you get a version of the same answer: the guest spends at the bar. That is true, and it is also the least examined number in the building. Very few properties have turned it into pesos per occupied bed, and most of the ones that have used the wrong figure to do it.

Everything below is measured on our own hostel, Maka, in Oaxaca, by the pricing engine that has rewritten its rate bands every morning since August. The pesos are ours. The arithmetic is what transfers.

Two rules before the division

The engine reads yesterday's bar and activity income every morning and divides it by the beds that were occupied. The division is trivial. What goes into the numerator is not, and two rules decide it.

Only named categories count as bar. Cocktails, beer, shots. Not a towel, not a lock, not a dorm bed booked late, not an Airbnb payout that lands in the same ledger, and above all not the catch-all line called Other. A standing check flags any income category that is neither bar, nor activities, nor a known incidental, which means a new or renamed category cannot quietly join the number or quietly leave it. If a bookkeeper has ever renamed a category in March and left you staring at a ratio that moved for no reason, you already know what that check is for.

Income is weighted by margin before it is divided. Our 2025 profit and loss puts bar margin at 27 percent and activities at 36 percent. A guest who runs a 100-peso tab does not hand you 100 pesos. He hands you 27. The rest is beer, a bartender, breakage and the fridge.

That second rule is where most back-of-the-envelope math breaks, and the error only ever runs one way. Use gross bar income and the extra guest appears to be worth 3.7 times what he is worth. That single substitution is enough to justify any discount you had already decided to make.

Weighted, divided, measured: our 2025 baseline is 32 pesos of ancillary margin per occupied bed. The engine's rolling average across the last eight logged days read 27.

Now put that next to a discount

Every morning the engine checks three neighbours on Hostelworld and comes back with a market floor. Across 32 logged runs that floor landed between MXN 168 and MXN 206. Round the bed to 200 pesos.

Take eight percent off it, which is exactly as deep as the engine's learned near-term discount ever went. You gave up 16 pesos of rate. Against that, the bed is carrying 27 pesos of measured ancillary margin. The discount looks paid for with 11 pesos to spare, and this is the argument that gets made at every hostelier dinner in this city.

It holds on one condition, and only one: that the discount is what put a body in that bed. Sixteen pesos given away on a bed that would have sold at full rate buys nothing at all, because the bar margin was walking through the door either way. The ancillary number does not defend a discount. It defends an incremental guest, and nothing else.

So we went and checked whether the guests were incremental

The engine deepened the near-term discount from 0 to -8 percent in three-point steps, chasing weak occupancy, and then compared what the two regimes produced. Days at -8 percent averaged 33.7 percent occupancy across 23 logged days. Days at 0 percent averaged 42.0 percent across 5. The deeper price did not bring more people through the door.

That comparison is weak and the code says so in its own comments: 23 days against 5, no day-of-week control, no seasonal control, all of it in low season. It is a direction, not a regression. But it is the direction that matters here, because the ancillary argument needs extra guests to exist, and on our data they did not. The engine logged the read and held its learned cap at 8 points rather than widening it toward the 25 points it is permitted to reach.

The number is noisier than you want it to be

Day by day, that margin per occupied bed ranged from 3 pesos to 123. A group that books a mezcal tasting lands on one date. A Tuesday in September with four people in the building lands on another. The engine will not report a rolling average at all until it has at least three days of data, and even then it moved from 34 to 27 across the last eight logged days.

Five pesos down on a series that swings by 120 is not a trend, and we are not treating it as one. It sits under the 32-peso baseline, which is worth watching and is not yet worth acting on. The useful discipline is knowing which of those two sentences you are standing in.

What to measure this week

• Your own margin rates, off your own profit and loss, line by line. Not an industry figure. Ours are 27 percent on bar and 36 percent on activities, and yours will not be the same.
• A fixed written list of what counts, plus one line in your monthly review that asks what income arrived in a category not on that list.
• Divide by occupied beds, not by guests, not by bookings, not by rooms. The bed is the unit you discount, so the bed is the unit the margin has to be expressed in.
• Three days minimum, read as a range. Write down the highest day and the lowest day, not only the average, so you can see how much of that average is one mezcal tasting.
• The incrementality question, in writing. Before the next discount: would this bed have sold anyway? If the answer is yes, the bar margin is not paying for anything.

What this does not claim

One owner-operated property, in Oaxaca, in low season. Every measurement here comes from runs logged between 14 August and 5 September 2026 at Maka Hostel, and the high-season behaviour of this mechanism has not been observed yet. The margin rates come from our own 2025 profit and loss and are ours alone. No revenue result is being claimed here: the finding is about how an occupied bed gets valued before a price moves, not about what any price move earned.

If this sounds like your month-end, the five checks above cost you an afternoon and fit in a spreadsheet with no software involved at all. If you would rather have it run every morning without you, see the three Nightfill tiers and start a free 14-day pilot on your own data: view pricing. Nightfill is operated independently of any channel or booking system, and results vary by property, market and season.