39. Tourism seasonality: what it is and how to measure it

In August you turn guests away; in January you pay rent and staff with the shutters half down. Anyone who runs a hotel, a restaurant, a visitor office or a local festival knows this swing. Tackling tourism seasonality means making that swing less steep: spreading visitor flows less tightly across the months (and across places), so that accommodation, services and jobs stay active for more of the year.

This guide explains what tourism seasonality is, how to measure it in a way that is comparable across countries, what official data show in Europe, Asia, Latin America and Oceania, where those data fall short, and what public policies actually do about it. Every figure comes with its source.

What is tourism seasonality?

In the indicator proposals for the statistical framework on the sustainability of tourism led by UN Tourism, the seasonality indicator "refers to variations in visitor arrivals on a regular time horizon and in regular frequencies". The document adds that the unequal distribution of tourist activity "is generally considered one of the major challenges for tourism businesses and local communities": in peak seasons, overcrowding, high prices and inadequate infrastructure; in shoulder and low seasons, too few services and too little work (UN Tourism, Measuring the Sustainability of Tourism: Proposals for a set of indicators, September 2023). The framework itself (SF-MST) was adopted by the UN Statistical Commission at its 55th session, 27 February to 1 March 2024 (UN Tourism, 5 March 2024).

Two clarifications prevent a lot of confusion:

• Seasonality is not a flaw in itself. Beaches in summer and snow in winter are real reasons to travel. The problem starts when concentration is so strong that services cannot cope in the peak and close down in the off-season.
• Statisticians use "seasonal adjustment" for something else. It is a method that removes recurring seasonal effects from a time series to show the underlying trend (Eurostat glossary). "Seasonally adjusted" figures hide exactly the pattern this guide is about, so check which kind of data you are reading.

What gets counted: nights, arrivals, visitors

Before measuring seasonality you need to decide what to count. International definitions come from the UN International Recommendations for Tourism Statistics 2008 (IRTS 2008) and, in Europe, from Eurostat's glossary:

Measure	Official definition	Why it matters for seasonality
Nights spent	Each night a guest, resident or non-resident, actually spends in a tourist accommodation establishment (Eurostat)	Eurostat's seasonality measures use it; it also reflects length of stay
Arrivals	A tourist arriving at a tourist accommodation establishment, i.e. a check-in (Eurostat)	Counts people, not days: a weekend and a week both count as one
Visitor, tourist, same-day visitor	A visitor travels outside their usual environment for less than a year; they are a tourist if the trip includes an overnight stay, a same-day visitor otherwise (IRTS 2008, paragraphs 2.9 and 2.13)	Accommodation statistics only see tourists who sleep there, not day trippers
Net occupancy rate of bed places	Nights spent in the period divided by bed places on offer, net of seasonal closures, times 100 (Eurostat)	Shows how full open capacity is; on its own it does not show how much closes in winter

The practical consequence: a one-day local festival that draws many people from nearby may leave no trace in nights spent. If you want to measure the effect of events, you need to know what your data source can and cannot see.

How do you measure seasonality in a way that compares across countries?

You need numbers that a mountain hotel, a coastal province and a whole country can calculate the same way. All three indicators below start from the twelve monthly figures for nights spent.

1. Share of the two peak months

The percentage of the year's nights that fall in the two busiest months. Eurostat uses it to compare European regions: "the share of the top 2 months in the annual number of nights spent", usually July and August (Eurostat, seasonality at regional level). UN Tourism's proposal applies the same idea to inbound arrivals, with peak months or the peak quarter (UN Tourism, 2023). With twelve identical months the share is 16.7% (2 out of 12); the higher it is, the more concentrated the year. In the southern hemisphere or in ski resorts the two peak months are different, which is why the indicator uses "top 2 months" rather than fixed months.

2. Seasonality Gini coefficient

Eurostat describes it as measuring the difference between the cumulative share of monthly nights spent, ranking the months from lowest to peak, and the distribution in the absence of seasonality, with annual nights spread evenly over twelve months (Eurostat). It is 0 when every month weighs the same; with twelve months the theoretical maximum is 11/12, about 0.917, when all nights fall in a single month.

The working formula, which applied to Eurostat data returns the published values (European Union 0.25 and Croatia 0.59 in 2025, checked by our editorial team on 1 October 2026), is:

G = Σ (2i - n - 1) × x_i / (n × Σ x), with months sorted from smallest (i = 1) to largest (i = 12) and n = 12.

This is the uncorrected form of the Gini, the one used throughout this guide. There is also a "corrected" form that multiplies the result by n/(n - 1), i.e. by 12/11, so that the maximum becomes 1. It is legitimate, but it gives higher values (0.442 instead of 0.405 in the example below) and does not match the figures Eurostat publishes. Before comparing two Gini values, check that they were calculated the same way.

3. Peak-to-average ratio

The busiest month divided by the average month. It is the easiest indicator to explain in a meeting: "in August we work 2.8 times a normal month". It is not an official Eurostat indicator, but it comes from the same data.

4. Share of the three peak months

The same idea as the two-month share, widened to the three busiest months. The OECD uses it in its 2026 report to compare member countries: "tourist nights in top three months as a share of total tourist nights", with 2024 data (OECD, Tourism Trends and Policies 2026). With twelve identical months it is 25%. According to the report, around 40% of tourist nights across OECD countries fall in the top three months; among the least seasonal countries it lists Japan (28%), Mexico, the United States and Colombia (29%) and Korea (30%), while in Norway (59%), Greece (52%), Sweden (51%) and Slovenia (50%) more than half of inbound tourist nights are concentrated in three months.

5. Coefficient of variation

It tells you how far, on average, the months sit from the average month: the standard deviation of the twelve values divided by their mean. It is 0 with twelve identical months and has no fixed ceiling. There is a choice to declare: the standard deviation can be calculated dividing by 12 ("population" standard deviation) or by 11 ("sample" standard deviation). We divide by 12, because the twelve months are the whole year, not a sample of months. The other choice gives the same year a value about 4.4% higher (the square root of 12/11), so two coefficients calculated differently cannot be compared.

6. Empty months

This is the indicator closest to the question of whether to stay open. We call a month "empty" when it accounts for less than 1/24 of the year's nights, i.e. less than half of what it would weigh in a year without seasonality (1/12). The threshold is our own convention and we say so: it lets you count, in the same way in every country, how many months fall below the minimum. With 2025 Eurostat data the European Union as a whole has no empty months, Italy has one and Croatia seven (our calculation on tour_occ_nim).

The conventions used in this guide, in one table

Indicator	How we calculate it	Where it comes from
Share of the two peak months	Nights in the two busiest months divided by the year's nights. Main indicator	Eurostat
Share of the three peak months	Nights in the three busiest months divided by the year's nights. Secondary indicator	OECD
Gini	Uncorrected form, months sorted from weakest; ranges from 0 to 11/12	Eurostat
Coefficient of variation	Population standard deviation (divided by 12) divided by the mean	Declared convention
Peak-to-average	Busiest month divided by the monthly average	Descriptive, not official
Empty month	A month with less than 1/24 of the year's nights	Our convention

A worked example, step by step

Take an imaginary destination with these monthly nights (a teaching example, not real data): January 200, February 220, March 300, April 450, May 600, June 900, July 1,600, August 1,900, September 800, October 400, November 250, December 380.

Step 0 - check the data before calculating. You need twelve months of the same year, from the same source and with the same definition (here: nights spent in all tourist accommodation). If a month is missing, do not calculate the indicator: a month guessed by eye moves the result more than you might think. If the source also publishes a "seasonally adjusted" series, use the original one: the adjusted series has removed exactly the pattern you want to measure.

Step 1 - total and average. 200 + 220 + 300 + 450 + 600 + 900 + 1,600 + 1,900 + 800 + 400 + 250 + 380 = 8,000 nights. Monthly average: 8,000 / 12 = 666.7.

Step 2 - share of the two peak months. The two busiest months are August (1,900) and July (1,600): (1,900 + 1,600) / 8,000 = 3,500 / 8,000 = 43.75%, against 16.7% for a flat year.

Step 3 - share of the three peak months. Add the third busiest month, June (900): (1,900 + 1,600 + 900) / 8,000 = 4,400 / 8,000 = 55.0%, against 25% for a flat year.

Step 4 - peak-to-average. 1,900 / 666.7 = 2.85: in August the place works almost three times an average month.

Step 5 - Gini. Sort the months from smallest to largest, multiply each by its weight (2i - 13) and keep the cumulative share alongside, which you need for step 6:

i	Nights (sorted)	Weight 2i - 13	Product	Cumulative share
1	200	-11	-2,200	2.5%
2	220	-9	-1,980	5.2%
3	250	-7	-1,750	8.4%
4	300	-5	-1,500	12.1%
5	380	-3	-1,140	16.9%
6	400	-1	-400	21.9%
7	450	1	450	27.5%
8	600	3	1,800	35.0%
9	800	5	4,000	45.0%
10	900	7	6,300	56.2%
11	1,600	9	14,400	76.2%
12	1,900	11	20,900	100.0%

The products add up to 38,880: the negative rows make -8,970, the positive ones 47,850. The Gini is 38,880 / (12 × 8,000) = 38,880 / 96,000 = 0.405.

Step 6 - cross-check with the Lorenz curve. The cumulative share column is the year's Lorenz curve: it tells you what share of nights the weakest months add up to. The exact values are 2.5 + 5.25 + 8.375 + 12.125 + 16.875 + 21.875 + 27.5 + 35 + 45 + 56.25 + 76.25 + 100 = 407, i.e. 4.07 as a decimal. The Gini can also be obtained this way: G = 1 + 1/12 - (2/12) × 4.07 = 1.0833 - 0.6783 = 0.405. If the two methods give the same number, the arithmetic is right; if they do not, a month has usually been sorted in the wrong place. The same column shows what the indicator means: the six weakest months together account for 21.9% of nights, against the 50% they would have without seasonality.

Step 7 - coefficient of variation. For each month, take the distance from the average (for August: 1,900 - 666.7 = 1,233.3), square it and add up the twelve squares: 3,394,466.7. Divided by 12 that gives 282,872.2, the population variance; its square root, 531.9, is the standard deviation. The coefficient of variation is 531.9 / 666.7 = 0.80.

Step 8 - empty months. The threshold is 8,000 / 24 = 333.3 nights. Below it are January (200), February (220), March (300) and November (250): four empty months. December (380) is just above.

Reading the results together. A two-month share of 43.75%, a Gini of 0.405 and four empty months describe a place with a single strong season and an almost idle winter: more concentrated than Italy or France and less than Greece (table below). To decide whether an off-season event makes sense, the most useful number is the last one: how many months fall below the threshold, and whether accommodation and services are open in those months.

In a spreadsheet you need four columns: sorted values, the weight (2 × row - 13), their product and the cumulative share. The Gini is the sum of the products divided by 12 times the total; the peak-month shares can be read from the last rows of the sorted list.

What the indicator cannot tell you

Move 300 nights out of August: 150 to April and 150 to October. The total stays at 8,000, the two-peak-month share drops to 40.0% and the Gini to 0.371. Now instead add 300 new nights in October without taking any from August: total 8,300, two-peak-month share 42.2%, Gini 0.394. Both times the indicators go down, but in the first case there is not a single extra visitor, in the second there is. The Gini describes the shape of the year; it does not tell you whether demand is new or shifted, or whether an event caused it.

What the data show: Europe and Japan

The table compares several countries with the same indicators, calculated by our editorial team from Eurostat monthly data (table tour_occ_nim, all tourist accommodation, updated 1 October 2026) and, for Japan, from the official monthly data of the Japan Tourism Agency (confirmed 2024 values, published 27 June 2025).

Country	Year	July + August share	Gini	Peak-to-average
Croatia	2025	54.5%	0.587	3.38
Greece	2025	41.6%	0.489	2.61
France	2025	36.0%	0.291	2.32
Italy	2024	34.5%	0.302	2.17
European Union (27)	2025	31.2%	0.249	1.95
Spain	2025	27.6%	0.217	1.75
Germany	2025	24.1%	0.168	1.49
Japan	2024	18.7%	0.052	1.21

Three useful readings:

• In Europe, summer weighs heavily. Eurostat reports that in 2025 nearly one third of EU nights were spent in July and August: August alone 16% (501 million nights), July 15% (460 million). Campsites are the most concentrated: 63% of their annual nights in the third quarter of 2025, against 36% for hotels (Eurostat).
• Japan is almost flat. In 2024 the busiest month (August) accounted for 10.1% of nights and the weakest (January) for 6.9%. The Japanese and European surveys classify accommodation differently, so compare the shape of the curve, not the levels.
• Within a single country the gaps are huge. Across European regions the two-peak-month share ranges from 66.0% in Romania's South-East region to 19% in Bucharest-Ilfov; in Italy from 58.8% in Calabria to 21% in Lazio (Eurostat, 2024 data). In five regions the peak month is February, because of snow (including Tyrol, Salzburg and Vorarlberg in Austria); in Vienna it is December.

For a closer look at one national case, region by region and compared with Europe and Japan, read Is Italy more seasonal than the rest of Europe?.

Residents and international visitors keep different calendars

The same place can have two overlapping seasons. In Croatia in 2025, 69% of international nights fell in the third quarter, against 23% for Croatian residents travelling at home (Eurostat). In Italy, residents peak in August and international visitors in July; in the third quarter of 2025 international visitors accounted for 53.4% of nights (ISTAT, tourist flows Q3 2025, in Italian). For organizers, a month that is "empty" for one audience can be full for another.

Beyond Europe: the southern hemisphere and less seasonal countries

The indicators exist precisely to compare countries with different calendars. Here are three cases, calculated or read with the same method.

Argentina: summer in January and a second season in July

Argentina's statistics institute (INDEC) publishes monthly nights spent in hotels and similar establishments. In 2024 there were 44,465,704. The busiest month was January (5,680,116 nights, 12.8% of the year), followed by February (4,939,392): the southern summer. The weakest was May (2,361,519, 5.3%). July, with winter school holidays and snow, was the third busiest month of the year (4,176,108) (INDEC, Encuesta de ocupación hotelera, November 2025, table 5, in Spanish).

Australia: arrivals peak in December

The Australian Bureau of Statistics (ABS) counts monthly arrivals of international visitors on short trips, meaning less than a year. In 2025, in the original series, there were 8,937,390: the busiest month was December (1,036,660), followed by November (831,170); the weakest was May (611,180) (ABS, Overseas Arrivals and Departures, table 1).

Indicator	Argentina 2024 (nights)	Australia 2025 (international arrivals)	European Union 2025 (nights)
Two peak months	January and February	December and November	August and July
Share of the two peak months	23.9%	20.9%	31.2%
Share of the three peak months	33.3%	29.7%	42.2%
Gini	0.132	0.075	0.249
Empty months	none	none	none

Our calculations on INDEC, ABS and Eurostat data, with the conventions described above. Two lessons:

• Two seasons pull the Gini down. Argentina has a summer peak and a winter peak: nights are split between two periods and the Gini stays low even though January weighs a lot. This is a case where the Gini on its own makes the year look flatter than it is; checking which months are the peak months avoids the mistake. Residents and non-residents both peaked in January and February, but for non-residents the second strong block was November-December (956,827 and 979,117 nights) rather than July.
• Arrivals and nights are not the same thing. In Australia the count is of border arrivals, not hotel nights: the ABS states that these statistics report the number of international border crossings rather than the number of people, so someone who enters twice counts twice. And a visitor who arrives in December may still be sleeping in the country in January. Arrivals tell you when people set off to come; nights tell you when they are there. The two curves look alike but do not match, and they should not share a column without saying so.

United States, Mexico, Colombia, Korea: steadier years

Many countries outside the European Union have no monthly series of nights comparable with Eurostat's. In its 2026 report, however, the OECD compared its members using the share of the three peak months: Japan (28%), Mexico, the United States and Colombia (29%) and Korea (30%) are among the least seasonal, which the report links to a more diversified offer that depends less on weather, with business travel, urban tourism and cultural activities all year round. The same report warns that, without more granular sub-national data, these national averages may mask important local differences (OECD, 2026). A country with little seasonality can still have very seasonal coastal or mountain destinations.

Why spreading demand is harder than it looks

Public strategies mention seasonality all the time. In its 2024 report on tourism trends and policies, the OECD notes that considerable resources go into business tourism (meetings, incentives, conferences and exhibitions) "in part to tackle seasonality", that highly seasonal mountain and coastal destinations try to extend the season "for example through events and festivals", and that destination management organizations promote the geographic and seasonal spread of tourists with human and financial resources that are often constrained (OECD, Tourism Trends and Policies 2024).

The same report sets a clear limit: modifying the tourism offer to extend the season "is not viable for all destinations". It also points to a newer factor: warmer winters are shortening the ski season, while heatwaves may make traditional peak periods less attractive and encourage visits at other times (OECD, 2024).

Some stated goals cited by the OECD:

• Croatia: its Sustainable Tourism Development Strategy 2030 makes extending the season a core priority.
• Portugal: its Tourism Strategy 2027 aims to reduce its seasonality index from 36.3% to 33.5%.
• France: the Destination France plan aims to extend the tourism season towards "four seasons".
• Canada: securing more international events as a four-season driver of travel.

One point of method applies everywhere: more nights in October do not necessarily mean new nights. Part of it can be demand moving from August, as the worked example above shows. That is good news for anyone who turns guests away in summer and closes in winter, but it is not a promise of growth, and it has to be measured with an honest comparison.

The most cited levers, and what to check for each

Lever	What it involves	What to check first
Off-peak events and festivals	Giving people a reason to travel, with a fixed date, in weak months	Who actually comes (residents, day visitors, overnight tourists), whether accommodation is open that week, whether the event clashes with others
Business tourism and conferences	Meetings, incentives, conferences and fairs, which follow the working calendar	Rooms, transport and beds available in the chosen weeks
Markets with different calendars	Audiences from other countries or with other holidays	Data by country of residence: when they already come, not when we would like them to
Offer that does not depend on weather	Culture, food and wine, walking routes, wellness	That basic services (restaurants, transport, information) stay open
Spreading across places	Moving part of the flows from the busiest spots to nearby ones	That the alternative places have capacity and services

If you want to move to practice from the organizer's side, the guide Organizing an off-season event explains how to choose the date, check which services are open and measure what happened afterwards.

Common mistakes when talking about seasonality

1. Comparing years with different definitions. In Italy, since January 2025, statistics also include private rentals managed on a non-business basis; ISTAT warns that this creates a discontinuity in the other-accommodation series and in the totals (ISTAT, monthly dataflow description). In Eurostat data, nights in Italy jump from 466 million in 2024 to 535 million in 2025: read that jump with this warning, not as growth. Check your own national statistics office for similar breaks.
2. Relying on one indicator. Cyprus had a higher Gini than France in 2025 (0.340 against 0.291) but a lower July and August share (29.6% against 36.0%): its season is long but its winter is very weak (our calculation on Eurostat tour_occ_nim). Read the indicators together.
3. Forgetting the reference year. Official data come out months later. Every table should say which year it covers and when it was read.
4. Using the pandemic years as a baseline. 2020 and 2021 are anomalous for any comparison.
5. Attributing a change to an event without comparing with a similar period or place without the event.

The limits of the data: what the numbers cannot see

Every indicator in this guide is only as good as the data it starts from. Before using a figure in a meeting, a plan or a funding application, check these seven points.

1. Who is inside the survey and who is not. In the European Union, countries may exclude establishments with fewer than ten bed places, and countries accounting for less than 1% of EU nights may raise the threshold to twenty (Eurostat, tourist accommodation reference metadata). In a small destination made of holiday homes and guest rooms, part of the supply may be missing.
2. People who do not sleep in tourist accommodation. Day visitors, guests of friends and family, second-home owners: accommodation statistics do not see them. For many local events they are most of the audience.
3. Arrivals, nights, border crossings. Three different measures, as the Australian case shows. Always compare the same measure, and say which one.
4. Original or seasonally adjusted series. Many institutes publish the same series in three versions: original, seasonally adjusted and trend (INDEC puts them in three columns side by side). To measure seasonality you need only the original series: the other two have removed exactly the monthly pattern.
5. Provisional data and revisions. The latest month published is often provisional: Chile's statistics institute says so in a note to every bulletin (INE Chile, December 2025 bulletin, in Spanish). Eurostat asks countries for monthly data within six weeks of the end of the month (Eurostat): a full year can be calculated mid-February of the following year at the earliest, and some figures may still change.
6. Series that change or stop. In Italy, since 2025, statistics also include private non-business rentals (see the first common mistake above). In Argentina, INDEC announced in January 2026 that its report on nights for November 2025 is the last under the current scheme: with the end of funding from the national tourism authority, the survey will be redesigned and its release rescheduled (INDEC, January 2026). Anyone following a series over time needs to know that a jump can come from the method, not from tourism.
7. Different classifications across countries. The Japanese and European surveys classify accommodation differently, and the same is true of almost any pair of countries: compare the shape of the curve, not the levels.

Three things remain that no seasonality indicator tells you: whether off-season demand is new or shifted (the worked example shows why), what the peak does to residents' daily lives, and what it costs those who run services. That is why UN Tourism places seasonality within a wider set of economic, social and environmental indicators.

For the technical terms used here (nights, arrivals, average length of stay, occupancy rate, shoulder months) see the tourism seasonality glossary.

Sources

All sources were consulted on 1 October 2026. Calculations on Eurostat, Japan Tourism Agency, INDEC and ABS data are by the Play the Event editorial team, using the method described in the text.

1. Measuring the Sustainability of Tourism (MST): Proposals for a set of indicators - UN Tourism (then UNWTO), 2023 (September). Consulted on 1 October 2026.
2. UN Adopts a New Global Standard to Measure the Sustainability of Tourism - UN Tourism, 2024 (5 March). Consulted on 1 October 2026.
3. International Recommendations for Tourism Statistics 2008 - United Nations and UN Tourism (UNWTO), 2010. Consulted on 1 October 2026.
4. OECD Tourism Trends and Policies 2024 - OECD, 2024. Consulted on 1 October 2026.
5. Seasonality in the tourist accommodation sector - Eurostat, Statistics Explained, 2026 (data extracted May 2026). Consulted on 1 October 2026.
6. Tourism statistics - seasonality at regional level - Eurostat, Statistics Explained, 2025 (2024 data, extracted November 2025). Consulted on 1 October 2026.
7. Nights spent at tourist accommodation establishments - monthly data (tour_occ_nim) - Eurostat, updated 1 October 2026. Consulted on 1 October 2026.
8. Glossary: Nights spent - Eurostat, Statistics Explained, consulted 2026. Consulted on 1 October 2026.
9. Glossary: Arrival of tourist at a tourist accommodation establishment - Eurostat, Statistics Explained, consulted 2026. Consulted on 1 October 2026.
10. Glossary: Net occupancy rate - Eurostat, Statistics Explained, consulted 2026. Consulted on 1 October 2026.
11. Glossary: Seasonal adjustment - Eurostat, Statistics Explained, consulted 2026. Consulted on 1 October 2026.
12. 宿泊旅行統計調査 (2024年・年間値（確定値）) - Overnight Travel Statistics Survey, 2024 annual confirmed values - Japan Tourism Agency (観光庁), 2025 (27 June). Consulted on 1 October 2026.
13. Movimento dei clienti negli esercizi ricettivi per tipo di esercizio - mensili (dataflow 122_54_DF_DCSC_TUR_3, descrizione) - ISTAT (Italian National Institute of Statistics), consulted 2026. Consulted on 1 October 2026.
14. Flussi turistici - III trimestre 2025 - ISTAT (Italian National Institute of Statistics), 2025 (3 December). Consulted on 1 October 2026.
15. OECD Tourism Trends and Policies 2026 - OECD, 2026 (approved 17 June 2026). Consulted on 1 October 2026.
16. Encuesta de ocupación hotelera. Noviembre de 2025 (hotel occupancy survey, November 2025) - INDEC (National Institute of Statistics and Censuses, Argentina), 2026 (20 January). Consulted on 1 October 2026.
17. Overseas Arrivals and Departures, Australia, July 2026 (Table 1: Total movement, arrivals) - Australian Bureau of Statistics, 2026 (11 September). Consulted on 1 October 2026.
18. Encuesta Mensual de Alojamiento Turístico (EMAT), diciembre 2025 (monthly tourist accommodation survey, December 2025) - INE (National Statistics Institute, Chile), 2026 (2 February). Consulted on 1 October 2026.
19. Tourist accommodation (tour_occ) - Reference metadata - Eurostat, 2024 (29 January). Consulted on 1 October 2026.