Methodology & data
How our crowd forecasts are built — and where the limits are.
What the score means
Every place gets a crowd score from 0 to 100. Lower is quieter. The score blends how busy a place tends to be at a given month, day of the week and hour, relative to its own typical range — not an absolute head-count.
What goes into it
The model combines seasonal visitation patterns, day-of-week rhythm by attraction type, the daily arc from opening to close, public holidays and school breaks, major events, cruise-ship arrivals for port destinations, and weather seasonality.
How current it is
These are modelled estimates grounded in well-established seasonality, not a live sensor feed. Each forecast shows the date its model was last reviewed and a confidence level so you always know how much weight to give it.
Where it can be wrong
One-off events, strikes, closures, weather swings and sudden changes in capacity can all move a real day away from the estimate. When we are less sure, we say so, and we separate what we observe from what we recommend.
Weather and events
For destinations added from October 2026, highs, lows and rain days are 1995–2024 averages from the nearest weather station in NOAA’s GHCN-Daily archive, or from the ERA5 reanalysis where no station has usable records; each page names its source. Events and holidays link to the organiser or tourism board that publishes them. Movable dates are marked as such.
Crowd index from overnight stays
For destinations added from 13 October 2026, the monthly crowd index is not set by hand. It is calculated from official monthly overnight stays (Eurostat for capital regions, or the national statistics office where it publishes city figures), averaged over 2022–2025. Each month is compared with the place’s own yearly average and converted to the 0–100 scale with a formula fitted on 18 destinations already on the site; the typical error of that fit is about 9 points. Overnight stays do not count day visitors or cruise passengers, so these scores carry medium confidence, and each page names the series it uses.
Confidence levels
Well-documented, stable seasonality with strong agreement across sources.
Clear overall pattern, with more variation from year to year.
Newer coverage — treat as a first estimate and expect refinement.
Free to use, no sign-up
Every forecast on this site is open to read in all six languages. There is no account, no paywall and no mailing list — you can just look something up and get on with your trip.
How we stay independent
The quietest option is always the one we show first, and no score is ever moved by anything other than the data behind it.
Spot something off?
Crowds change, and so should our model. If a forecast doesn’t match what you saw on the ground, tell us and we’ll review it.
Report a correction