ES EN
Back to blog
Nightfill blog

We read the same six cities twice. The floor held and the median moved 33 per cent.

Six cities read on 6 or 7 October and again on 12 October. The cheapest bed moved under 2.1% in five of them. One median moved 32.7%.

Published October 13, 2026 · 6 min read
Charming Andalusian courtyard with lush plants in Cordova, Spain, capturing traditional Spanish architecture.
Photo: Domenico Adornato / Pexels

Six cities, read twice

Every morning our engine opens the public Hostelworld city page for a city on our prospect list, takes the listings on it, and records what they were asking for a dorm bed on the sampled night. Nothing below is modelled or estimated. It is what the pages said when we looked.

The pass on the morning of 12 October, Mexico City time, covered six cities, and every one had already been read once, on 6 or 7 October. That is the first time we have had the same city on two separate mornings, and the pair is worth more than either alone. Both readings below, as cheapest / median / dearest, in USD per bed per night for one sampled night, from the listings carrying a price when we looked.

  • Playa del Carmen, 7 of 30 priced both times. 6 Oct, night of 9 Oct: USD 10.48 / 13.09 / 22.63. 12 Oct, night of 16 Oct: 10.33 / 13.49 / 22.30.
  • Cartagena, 7 of 30 both times. 6 Oct, night of 9 Oct: 11.71 / 23.75 / 29.51. 12 Oct, night of 16 Oct: 13.90 / 15.98 / 28.21.
  • Lima, 6 priced both times. 6 Oct, night of 9 Oct: 10.31 / 12.81 / 21.81. 12 Oct, night of 16 Oct: 10.40 / 13.44 / 16.13.
  • Cuzco, 7 of 30 both times. 7 Oct, night of 10 Oct: 8.60 / 13.59 / 26.86. 12 Oct, night of 4 Nov: 8.66 / 13.20 / 16.97.
  • Arequipa, 6 of 30 both times. 7 Oct, night of 10 Oct: 8.90 / 10.14 / 12.05. 12 Oct, night of 17 Oct: 8.90 / 9.94 / 12.15.
  • Tulum, the only pair read for the same night, 17 Oct. 7 Oct: 4 of 31 priced, 8.59 / 9.03 / 28.75. 12 Oct: 5 of 30 priced, 8.41 / 10.34 / 25.30.

The floor barely moved. The middle and the top did.

Side by side, the three behave nothing like each other.

The cheapest bed, five or six days apart: Arequipa 8.90 both times, to the cent. Cuzco up 0.70 per cent, Lima up 0.87, Playa del Carmen down 1.43, Tulum down 2.10. Five of the six moved under 2.1 per cent. The sixth, Cartagena, went up 18.70 per cent, from 11.71 to 13.90.

The median: Arequipa down 1.97 per cent, Cuzco down 2.87, Playa del Carmen up 3.06, Lima up 4.92, Tulum up 14.51, and Cartagena down 32.72, from 23.75 to 15.98.

The dearest bed: Arequipa up 0.83 per cent, Playa del Carmen down 1.46, Cartagena down 4.41, Tulum down 12.00, Lima down 26.04, and Cuzco down 36.82, from 26.86 to 16.97.

That inverts what a rate-shopping dashboard trains you to expect. The number every tool sets in the biggest font is the market median, the least stable of the three. The number nobody plots is the floor, and in five of six cities it was the same number twice.

Why a median built on seven prices moves on its own

Before reading any of that as a market, read it as a sample. These medians are built on six or seven prices, not thirty: the city page shows roughly the first thirty listings, and our engine prices the one it matched plus the six beside it. The median is the middle of six or seven asking prices, in the order the channel chose to show them.

Change which seven priced that morning and the median changes without a single hostel touching a rate. Tulum is the clearest case, the only pair read for one night. On 7 October four listings carried a price; on 12 October, five. With four prices the median is the average of the second and third, which on 7 October added up to 18.06. With five prices it is the third price outright. The reported median went from 9.03 to 10.34, up 14.51 per cent, and the arrival of one more priced listing is enough to produce all of it.

Cartagena deserves the same suspicion. Its median fell 7.77 while its cheapest bed went up 2.19 and its dearest fell only 1.30: the ends held and the middle collapsed. The gap from cheapest to median was 12.04 on 6 October and 2.08 on 12 October. A market where the floor rises and the top is flat did not fall by a third. Far more likely, a cluster of mid-priced beds was in the readable seven on one morning and not on the other.

What a 25-bed house does with this

Take 25 beds running 60 per cent: 15 beds a night, 450 bed-nights in 30 days. Price the movement between the two readings as though you had followed it.

Cartagena's median swung 7.77: on 15 beds a night, USD 116.55 a night and USD 3,496.50 across 30 nights. Cuzco's dearest swung 9.89: USD 148.35 a night, USD 4,450.50 across 30. Those are the two numbers that look like signal, and both may be the sample changing.

Playa del Carmen's floor swung 0.15, on the same 15 beds USD 2.25 a night and USD 67.50 across 30 nights. Arequipa's swung nothing at all. The stable number is the one worth almost nothing to track, which is the point: a number that does not move is one you can anchor to.

The move is smaller than the noise

Before changing a rate off a competitor reading, write down the move you are considering, then how much that number moved between two mornings when you did nothing.

A 5 per cent move on a Cartagena bed at the 12 October median of 15.98 is 80 cents. The median itself moved 7.77 between the two readings, 9.7 times the move you are planning. Cut 80 cents on Monday, and if the median has moved several dollars by Thursday the reading cannot tell you whether the cut worked. You get a number back and it means nothing, which is worse than having no number, because you will act on it.

So the working rule is short. Price against the floor and your distance from it, because that is the number that stayed still. Treat one morning's median as a sketch of the market's shape, not a level. And before moving on any competitor number, read the same city for the same night twice, and believe only the part that shows up in both. Nightfill takes that reading every morning and makes the pricing decisions off it on a property's behalf; the three tiers are at view pricing. It is independently operated, and what it does depends on your property, your market and your season.

The same read, for your own hostel

You do not have to buy anything to check this on your own city. Paste your Hostelworld listing at nightfill.co/#read with an email address, and the same reading arrives for your property by email in about ten minutes: your cheapest dorm bed against the other hostels on your city page, night by night, your rank from cheapest, and the nights you sit above your city's median. Then do what this post is about and read it again next week, before you move anything.

revenue-managementpricinghostelworldrate-shoppingdatosmedicion