Methodology: how HolidayHub calculates crowding

“How busy is it?” has two different answers, depending on what you compare against. We calculate both and never merge them into a single value.

One rule follows from this, and we hold to it in every publication: across travel regions we compare only on density. The score may be used across regions only with the explicit qualifier “measured against its own annual pattern”. “Region A is emptier than region B” is not a valid statement on the score.

In the interface the score is given a word: below 34 the week counts as quiet, from 34 as average, from 67 as peak season. The exact wording differs by view. One exception takes precedence: if the busiest week is no more than 20% above the annual mean, it reads “little seasonal variation”. A flat annual curve should not be called peak season merely on the strength of its relative maximum.

The starting point is the official accommodation statistics of the European statistical regions. We average them over several years rather than take a single one, so that an exceptional year does not shape the seasonal curve. Within a month, the holiday calendar of the European source markets shifts the weight onto the holiday weeks; how strongly is a reasoned assumption, not a measured value. The result is available per calendar week: once as a score against the region's own annual pattern, once as a density per square kilometre.

The model is laid out in four layers. The table shows what is in the published score of 06/08/2026, not what exists in the code:

Source-market weighting only applies where the official statistics report which countries the guests come from; for the remaining travel regions every covered country counts equally. The figure above is measured, not estimated.

The limits belong with the figure, so they stand here and not in a footnote.

• It does not measure, it models. We count no people on the ground and read no real-time occupancy. The figure says how busy a week is likely to become given the official data.
• Day visitors are missing. In the EU-27 in 2024, some 1.67 billion day trips stood against only 1.19 billion overnight trips. Eurostat itself writes that the statistics are likely to underestimate the true extent of tourism pressures.
• Cruise passengers are excluded from the survey, and establishments with fewer than 10 beds are not recorded. Destinations with heavy day or cruise tourism (Venice, Dubrovnik, Santorini) appear too empty rather than too busy here.
• Several travel regions can lie in the same statistical region; four Canary Islands do. They then share the same curve and the same density. The figure there is a statement about the statistical region, not about the individual island.
• Density spreads the overnight stays across the entire area of the statistical region. A busy place in a large, otherwise empty region therefore appears too empty.
• The score is an average over several years, not the calendar of a particular year. It knows neither the holiday pattern of a single year nor major events.
• The climate figures shown alongside crowding rest on windows of differing length: precipitation and snow on only 3 years. They belong to the weather comfort score, not to crowding.

We cover 288 travel regions in Europe. For 282 of them the official data is sufficient for a weekly crowding figure; for the remainder we show climate and accessibility but no crowding. A named gap is more honest than a guessed value. Public holidays cover 42 countries, school holidays cover 35. Only school holidays modulate crowding.

The index carries a version. If the calculation changes (a different formula, a different weighting, an additional layer), the version number rises and we record the step here; the previous definition stays documented so that an older citation remains traceable. If only new data arrives, the data date alone changes.