We pulled two years of Wikipedia pageviews for the tourism article of 183 countries: 2.07 million reads, median down 16.6% in twelve months, twice the decline of Wikipedia itself. Only 22 countries gained attention. Almost none of the winners are where the value is.
Wikipedia keeps a public ledger of what the world is curious about, one pageview at a time. We pulled it for travel: monthly views of the English "Tourism in ..." article for 183 of the 199 countries we rate, September 2024 through August 2026, via the Wikimedia Pageviews API. That is 2.07 million human reads in the last twelve months, compared year over year, then crossed with our cost floors and seasonal ratings.
Two findings up front. Travel reading is in a measured recession: the median country lost 16.6% of its tourism-article readers in a year, more than double the 7% decline of English Wikipedia as a whole over the same windows. And the countries gaining attention have almost nothing to do with the countries gaining value.
The recession first
The Wikimedia Foundation itself reported human pageviews down about 8% last October and named the cause: AI assistants and search summaries answering questions that used to be clicks. Our slice says travel questions are being answered that way at twice the general rate, which tracks: "when should I visit Portugal" is exactly the query a chatbot swallows whole.
Of 156 countries with enough prior-year traffic to compare, 22 gained readers. 29 lost more than 30%. Portugal (−49%), Pakistan (−47%), India (−47%), Argentina (−46%) and the Maldives (−42%) shed nearly half their audience without any corresponding collapse in actual tourism. The reading moved; the traveling didn't. You are, for the record, reading a travel site's argument that nobody reads travel pages anymore. We see the irony, and we also see our server logs agreeing with it.
What made 22 countries swim against that tide
Almost every riser decodes to a single dated event, visible as a spike in the monthly series.
| Country | YoY | The event in the data |
|---|---|---|
| Cape Verde | +128% | ~530 reads/month baseline; 5,500/month in June and July 2026, its first World Cup, where it reached the knockouts; back to 780 by August |
| Bahamas | +99% | no single clean event; the ledger occasionally shrugs |
| Rwanda | +68% | rise clusters in late 2025, the quarter Kigali hosted the first UCI Road World Championships in Africa |
| Venezuela | +64% | one January 2026 news spike; 372 reads this August. Headlines, not holidays |
| Kyrgyzstan | +25% | steady climb, no spike: the rare organic riser |
| Croatia | +13% | broad, sustained; also the rare organic one |
| United Kingdom | +11% | peaks at 4,664 in April 2026, the month its ETA price hit £20 |
| Laos | +11% | gradual, consistent with the China rail-link tourism build |
The pattern worth keeping: a World Cup or a championship buys a country two months of attention and then the series reverts to baseline. Attention spikes decay; they are not demand.
The most-read tourism article is for the country you cannot visit
Tourism in North Korea: 92,838 reads in twelve months, the largest in the ledger, ahead of the United States (80,194), Thailand (77,464) and Japan (74,722). Roughly 7,000 people a month read about tourism in a country that admits a few thousand Western tourists a year. A chunk of Wikipedia travel reading is entertainment wearing a planning costume, which is worth remembering before anyone sells you "search interest" as a demand signal.
Attention and value are strangers
Price does not explain who gets read: countries with floors under $15 a day lost a median 15.9% of readers, countries over $40 lost 16.9%. Effectively identical. The internet's attention is distributed by news cycle and name recognition, not by what anything costs.
Which creates the arbitrage this site exists for. Sri Lanka lost 40% of its readers in the same year its currency made it 9% cheaper for euro and dollar travelers, with a $7-a-day floor and its south-coast season opening in November. The attention ledger says forgotten; the value ledger says first pick.
The wider screen — floor under $17, rated "best" somewhere in October–December, under 20,000 readers a year, audience not collapsing — returns Benin ($9, +6.7%), Guyana ($15, flat), Ivory Coast ($17, +18%), Nicaragua ($12), and Moldova ($15). Two caveats belong in the same sentence: several West African entries carry regional travel advisories the pageview data knows nothing about, and Colombia would top this list on attention but its peso just made it 23% pricier. Check the country guide before the flight search, not after.
Method
Wikimedia Pageviews REST API, per-article monthly series, en.wikipedia, all-access, agent class "user" (bot traffic excluded by Wikimedia's classifier). One "Tourism in X" article per country; 16 microstates have none. YoY compares Sep 2025–Aug 2026 against Sep 2024–Aug 2025; countries under 600 prior-year views are reported but not ranked. Known limits: English-only skews toward anglophone interest; article-title redirects fragment counts for a few countries; and one article per country is a proxy for attention, not a census of it. The control (−7.0%) is the same computation on total en.wikipedia user pageviews. Raw series and scripts are documented in the seed file for this post, and the when-to-visit ratings it crosses were themselves audited against 363,374 days of weather data last month.