Hotel Occupancy 2026: Five Destinations Fall, Cozumel Rises

Five of the six comparable Mexican Caribbean destinations have now posted three consecutive years of falling hotel occupancy. Cozumel, over the same window and in the same table, is rising. Both facts are published on the same page of the same official bulletin — and that contrast, not the decline, is the part worth looking at.

This article does not compare figures across different bulletins, and it derives no monthly number. Everything below comes from a single PDF, and where it does not, it is labelled.

What does this figure actually measure?

Before any number, what the number is. The bulletin calls the column «Ocupación», and it measures this:

• Hotel occupancy, built from average daily available rooms and occupied rooms. It excludes Airbnb and every short-term rental platform. These are registered hotel establishments only.
• Year to date, 1 January to 30 June — not the month of June on its own.
• Preliminary results. That is how the document titles them, in its own header.
• Published by Mexico’s National Tourism Statistics Information System (Datatur, Ministry of Tourism) in its monitoring bulletin for the period ending June 2026.

The full document, a publicly accessible PDF in Spanish: hotel monitoring bulletin, year to date through June 2026 (Datatur / Sectur). Every figure in this article marked [D] comes from it.

So every figure can be traced without leaving the text, they are marked:

• [D] — published by Datatur, copied verbatim.
• [N] — our own arithmetic on published figures, with the method shown.
• [A] — AirROI, a private source measuring something else.

How far did occupancy fall in each destination?

The three year columns cover the same calendar window: 1 January to 30 June of 2024, 2025 and 2026 respectively. The bulletin defines them that way at the foot of the table. That is what makes them comparable side by side.

This is not a Caribbean-wide decline

The easy reading of this table is «the Mexican Caribbean is down». The table does not say that. It says five destinations have strung together three years of decline and one has strung together three years of growth — same methodology, same period, same document [D].

And the magnitudes are nothing alike. Akumal loses 19.3 points over two years; Cancun loses 3.5. Same state, same window, a difference of more than fivefold [N]. Folding them into one phrase — «the area», «the Caribbean», «the Riviera» — erases precisely the fact that matters.

Is the country falling, or is the Riviera Maya?

This is the question almost nobody asks, and the bulletin answers it two rows higher, in the same columns.

Mexico’s beach-destination average is also down: 3.9 points over two years [D]. There is a national tide, and it is real. But the Riviera Maya loses 9.8 points over that same window — roughly two and a half times the national beach average — and Akumal nearly five times [N].

That is the difference between quoting a number and understanding it. A destination falling less than its own national average is gaining ground in relative terms even as its figure drops; one falling two and a half times faster is losing ground even though the tide explains part of it. The Riviera Maya’s decline is steeper than the country’s, and that cannot be pinned on a national cycle.

The bulletin adds one caveat about these aggregates that is worth honouring: «Occupancy by type of destination and the total are a weighted average. Variations may not match due to rounding» [D].

Cozumel, the one going the other way

Cozumel moves from 56.3 to 58.1 to 59.3 [D]. It rises in all three years, and it is the only destination on the list doing so while its own national benchmark falls.

It is worth saying what this figure does not say. Cozumel starts from the lowest base of the six: its 59.3 in 2026 still sits below Akumal’s 63.5, the destination that fell hardest. Climbing three points from 56 and dropping nineteen from 83 are not equivalent stories, and neither is explained inside the bulletin.

This is where we stop. We do not know why Cozumel is rising. The document publishes available rooms, occupied rooms and occupancy; it does not publish causes, and we are not going to supply them.

Why is Isla Mujeres not in the comparison?

Quintana Roo has seven monitored destinations in this bulletin, not six. The seventh is Isla Mujeres, and its figures are 55.2, 57.7 and 50.4 [D].

We left it out of the table above, and the reason is not ours. Isla Mujeres is the only one of the seven carrying a footnote marker in the bulletin. In the original Spanish, it reads:

«Apartir del primero de septiembre 2025 se actualizo la oferta hotelera del centro por lo que los datos no son estrictamente comparables con periodos anteriores.»

— As of 1 September 2025 the destination’s hotel inventory was updated, so the data are not strictly comparable with earlier periods.

In other words: the source itself warns that year-on-year comparison does not hold there. We publish the figure because it exists and is public, and we keep it out of the comparison because Datatur, not Nautilus, signed that exclusion. The other six destinations carry no marker at all.

Why does Tulum appear in no table here?

Datatur does not monitor Tulum. The figure is not low, and it is not buried: it does not exist in the bulletin. The Quintana Roo destinations it publishes are Cancun, Cozumel, Isla Mujeres and Riviera Maya — with Akumal, Playa del Carmen and Playacar as sub-destinations of the last.

Any «Tulum hotel occupancy» figure citing this bulletin is citing something the bulletin does not contain. The Tulum figure we do use across the site is a different metric from a different source: 28.9% median Airbnb occupancy, AirROI, June 2025 to May 2026 [A]. It is private, it measures platforms rather than hotels, and for that reason it cannot sit in the same table as the numbers above. It appears in our Tulum vacation rental guide and our Tulum zone guide, always with that label attached.

What this bulletin cannot tell you

Four gaps, stated out loud because they are part of the data:

• It does not say why occupancy is falling. The bulletin publishes rooms and occupancy. New supply, sargassum, exchange rates, air connectivity — none of that is in the document, and attributing it would be invention.
• It does not say whether the decline has a floor. Three observations pointing the same way are three observations, not a trajectory.
• It does not allow seasonality to be separated from trend. Datatur publishes the year-to-date figure, never the isolated month. We tried deriving the month by subtracting consecutive bulletins and the result does not hold, for the reason in the next point. Comparing January–June against January–June removes the seasonal problem, but it also makes those six months opaque from the inside.
• The history moves. Datatur restates closed periods: each bulletin recomposes the year-to-date figure with whichever hotels reported. Comparing the same January–June of one year across two bulletins published at different times, we found gaps of up to 10.6 points [N]. That is why every figure here comes from one bulletin: mixing publications mixes methodologies, and produces numbers that look like data without being data.

How we read an occupancy figure

This is not a forecast or a recommendation to buy. It is the standard we apply to this kind of figure when it turns up in a sales deck:

• We ask what it measures before what it says. Hotel occupancy and Airbnb occupancy are different numbers sharing a name. A brochure that blends them is not exaggerating — it is measuring two things and calling them one.
• A destination’s occupancy is not a unit’s. A destination average pools hotels of every category; it does not describe one apartment, with its building, its management and its location.
• We ask for the document, not the number. An occupancy figure with no bulletin, period and issuing body behind it cannot be verified, and what cannot be verified does not enter a valuation.
• We compare within a single document. If a source restates what it already published, two figures from two bulletins are not comparable even when both are official.

Nautilus does not publish appreciation, return or occupancy projections. What we do publish, with its source and date attached, lives on our market data page.