The city has 106 and we price against three

A hundred and six, under two names

Our discovery job walks a list of cities twice a day and records every property in them that sells the kind of bed we sell. After the run at 00:05 Mexico City time this morning, 5 October, it holds 2,620 properties across 60 city labels. One of those cities is Oaxaca, where I run a 46-bed hostel. Oaxaca shows up twice in our own file: 93 properties under "Oaxaca" and 13 under "Oaxaca de Juarez". That is our labelling, not two towns. Call it 106.

The engine that rewrites our rates every morning prices against three of them.

Those two numbers next to each other are the subject of this post. Not because either is wrong, but because most operators I talk to carry a third number in their head, the four or five properties they call the competition, and have never checked it against either.

What the map says

Same file, this morning, for the markets we sell into:

• Oaxaca 106, Lima 52, Playa del Carmen 38, Medellin 35
• Arequipa 34, Buenos Aires 31, Bariloche 30, Cusco 29, Mendoza 28
• Mexico City 25, spread across six neighbourhood labels
• Santa Marta 22, Bogota 21, San Cristobal de las Casas 20, Cartagena 16
• Tulum 13, Quito 13, Puerto Escondido 12, Panama City 10, Montevideo 10
• Bocas del Toro 7, Cuenca 7, Puerto Viejo 5, La Fortuna 4, Mancora 3, Colonia del Sacramento 2

Two caveats before anyone quotes those. They come from contributor-made map data with our enrichment on top, so each is a floor and not a census: 1,194 of the 2,620 rows carry no city label and are in none of the counts above. And a property counts as one whether it holds 12 beds or 180. For the night you are filling, beds compete, not front doors.

What makes me trust the shape of it, if not the exact figure: this morning's run covered eight cities and returned 622 candidates, 620 of them already in the file. Two were new. A market we have mapped stays mapped, which is the one thing you want from a denominator.

Three is the instrument, not the market

Our engine reads the market each morning from three named properties on Hostelworld, on one fixed daily probe. Across 32 logged runs between 14 August and 5 September that reading came back between MXN 168 and MXN 206. A narrow, fixed spec is the right way to take that reading: it is what makes today's number comparable to yesterday's.

So three is not an error. Three is a gauge. The mistake is letting the gauge stand in for the market, and the 106 is what tells you how far apart those two things are. Three properties can tell you whether the floor moved. They cannot tell you whether a move of yours will be noticed, matched, or absorbed into noise. Density decides that, and density is the number almost nobody has written down.

What a hundred and six does to a discount

In a market with 106 comparable properties, your rate is not an announcement. A guest comparing beds in Oaxaca sees a screen of prices where yours is one line and the one under it is a near-identical dorm at a near-identical number. Cutting yours does not make you visible, and provokes no response, because 105 operators are not watching you.

We tested that on ourselves twice this season. Both readings point the same way.

The engine's own: 23 logged days with the near-term bands at minus 8 per cent averaged 33.7 per cent occupancy, against 42.0 per cent on the five days at zero. Deeper price, lower occupancy. It wrote that line in its log and held its discount cap instead of digging further.

Then the week we overrode it by hand. From 27 September we set the same-day and one-to-three-night bands to minus 50 per cent through 4 October. The beds did arrive, from night four onwards, and the week still took less money than the week before it at half the occupancy. Neither reading is a controlled experiment, and I said so where each first appeared. But in a 106-property market the thing that failed to happen both times is what operators expect: the cut did not buy volume worth the margin it gave away.

If your city is dense, the rate is not your lever

Twenty or more comparable properties and your price cannot do the work. What is left is what is yours alone:

• Lead time. A bed 30 days out and the same bed at four this afternoon carry different risk. Selling both at one price is the most common unforced error I see.
• Length of stay. One four-night booking costs you one changeover instead of four, and is worth more than its nightly rate says.
• Direct share. One real booking of ours netted out at 26.7 per cent effective commission once its promo stack was counted. Every booking you move to direct is that figure back in your pocket, with no guest paying more.
• Margin per occupied bed, not per bed sold. Ours runs at 27 pesos per occupied bed on bar and activities, against a 32-peso baseline from the 2025 accounts. That number decides whether a cheap bed is worth having, and it is invisible if you only look at the rate.

If your city is thin, assume you are read

Bocas del Toro has 7 in our file. Colonia del Sacramento has 2. At that density your rate is public information with an audience, and the owner down the road has your listing open in a tab. A cut there is visible within a day and matched within two, leaving everybody at the lower number holding the same beds. The thin-market discipline is the opposite: track all of them by name and compete on terms you can hold, not on a rate that will be copied by Thursday.

Counting your own, in an afternoon

• Open the channel that sends you volume. Set one date three days out, one guest, one night, and keep that spec.
• Count every listing selling the bed you sell within a fifteen-minute walk. Not the boutique hotel. The bed a guest would take instead of yours.
• Write the count down. Under 10 and you are being read. Over 20 and your price is noise.
• From that list pick three: same bed, same neighbourhood, same channel. What you keep from them is the band they live in, not today's number.

What this does not prove

These are counts of properties we could find, not of beds, and not of your market's true capacity. Our file carries two labels for Oaxaca and six for Mexico City, and where I added labels together I said so. 1,194 rows have no city at all. The two pricing readings come from one 46-bed property in one low season, with no day-of-week control. Take the method; measure your own numbers.

If this sounds like your property, see the three Nightfill tiers and start a free 14-day pilot on your own data: view pricing.