About & methodology

I went through 301 YouTube travel vlogs and wrote down every place the vloggers visit or recommend in 27 cities. Each city guide ranks those places by how many vlogs mention them and links every mention to the exact second in the video, so you can see the place for yourself before you go.

Every place comes from popular English-language YouTube travel vlogs: 301 of them from 140 channels so far. I only use a vlog's original English captions (YouTube's speech recognition of the spoken English, or the creator's own English captions), never machine-translated tracks.

• Vlogs are found by searching YouTube for the city's travel guides, itineraries and “things to do” videos; the title has to name the city.
• A title filter skips videos that rarely name places you can visit. Itineraries and guides are kept, also when they cover food.
• Pure food tours are mostly skipped: they rarely name restaurants or stalls you could find. One is used only when it names the places.
• Videos without spoken English are skipped, because machine-translated captions garble place names.

1. Transcript. The vlog's English captions, with a time for every line.
2. Place extraction. A large language model reads the transcript and lists every place that is visited or recommended, with its proper name, category, area and the moment it comes up.
3. Timestamp check. Each timestamp is checked against the transcript around it. If the place isn't discussed there, a stronger model reads the whole transcript and moves the mention to where it really is, or removes it when the place is only named in passing.
4. Duplicates merged. The same place under different spellings or names (common with speech recognition) becomes one entry, and garbled names are corrected.
5. Ranking. Places are ranked by how many vlogs mention them: the more vloggers who independently went there, the higher it sits. Ties go to the place more vlogs agree on, then to the most-watched vlog.

Place descriptions and notes are rewritten into plain, visitor-facing text; the rewrite may rephrase but never add facts that the vlogs don't contain.

“Mentioned in 5 vlogs” means five different vlogs show or recommend the place. A vlogger who comes back to the same place three times in one video still counts once. Some mentions are brief, a place named on the way to somewhere else; they count too, but the first clip a guide links to is the longest real segment about the place.

New vlogs are checked several times a day, and the guides are republished a few times a day when something changed. A new city goes live once it has enough vlogs and places for the ranking to mean something.

• English only. Places that mostly appear in vlogs in other languages are missing or rank too low.
• Vloggers favour tourist cores. Central neighbourhoods and famous sights are over-represented; quiet residential areas and places outside the centre are under-represented.
• A place shown by a single vlog is weaker evidence than one many vlogs agree on. Use those as leads, and check opening hours before you go: vlogs can be a few years old.

No. There are no ads, no sponsorships, no affiliate links and no hotel partnerships. Nobody can pay to be listed or to rank higher; the order comes only from the vlogs.

Suggested citation:

What to do in (whattodoin.fyi), “About & methodology”, updated October 18, 2026, https://whattodoin.fyi/about

For a city, cite its guide, for example https://whattodoin.fyi/tokyo. A plain-text version for AI assistants is at /llms.txt, with one markdown file per city at /llms/<city>.md (for example /llms/tokyo.md). You are welcome to quote the rankings with a link back.

Missing a city, or found a place that's wrong? Use the request form on the home page.