This case study explores the use of mobility data to measure domestic tourism movement in Australia. Traditional survey methods, such as the National Visitor Survey, have been used for over 40 years to collect data on domestic tourism. However, declining response rates and increasing costs have prompted Tourism Research Australia to investigate new data sources. The mobility data project, developed in partnership with DSpark, leverages mobile positioning data to provide more granular and timely reporting of domestic tourism estimates. This innovative approach combines mobility data with stratified household surveys, census data, and financial transactions data to enhance accuracy and reduce costs. The project aims to improve the measurement of domestic overnight trips, daytrips, and overall tourism movement, providing valuable insights for investment and infrastructure development.
Using mobility data to measure domestic tourism movement in Australia
Abstract
Australia has been measuring both international and domestic tourism using traditional survey methods via large-scale sample surveys for over 40 years. Tourism Research Australia is part of the Australian Trade and Investment Commission and is the government body responsible for managing Australia’s national surveys, the International Visitor Survey and National Visitor Survey. These surveys have been supported through Commonwealth and state co-funding agreements during this time.
International visitation to Australia accounts for 30% of total tourism spend and provides jobs, business opportunities in investment and trade. Domestic tourism represents approximately 70% of the total spend for tourism in Australia. It is the major contributor to Australia’s regions and is a key driver of investment and infrastructure.
The International Visitor Survey and National Visitor Survey were customised for tourism and follow the International Recommendations for Tourism Statistics. The survey estimates have provided quality, consistent data over several decades and are core inputs to Australia’s Tourism Satellite Account, balance of payments and trade in services statistics.
This case study focuses on domestic tourism. Measuring domestic tourism has been undertaken for almost 40 years using computer assisted telephone interviewing. The National Visitor Survey has been collected from 120 000 interviews each year. Results are benchmarked using population information as published by the Australian Bureau of Statistics. This mode of collection was favoured for its economic benefits as it was much cheaper than the optimal approach of using stratified household surveys. The National Visitor Survey collects a broad range of information on Australian resident travel including domestic overnight, daytrips and Australians travelling internationally.
The mobility data project
Tourism Research Australia has been investigating new data sources for use in tourism statistics for over a decade. The mobility data project has been underway for over 3 years, coming to completion in 2023. This mobile positioning data is one of a range of data sources that form a modelled solution for declining survey response rates.
For the purposes of this case study, the focus is on the mobile positioning data aspect of the project in relation to domestic tourism. While work is underway into the alternatives for international data as well, it is less advanced due to impacts of COVID-19 on the international data series. There are also additional complications related to measuring international travel by means other than tailored surveys.
Work is also well underway in determining if financial transactions data are fit for purpose for tourism statistics. As with the use of mobility data for movement, use of financial transactions for domestic tourism statistics is more advanced than for international tourism statistics.
Rationale
Maintaining a reliance on large scale sample surveys has become increasingly difficult. Falling response rates are causing declining data quality, increasing collection mode bias and increasing labour costs It is increasingly difficult to provide a representative sample across the population, with higher income households more likely to participate in telephone surveys. The Pew Research Centre in the United States reports that telephone response rates have fallen as low as 6%. Tourism Research Australia has seen the response rate for the National Visitor Survey fall from 50% in 2016 to 15% by July 2022 (Figure 1). This trend has continued with rates in late 2023 at 10%-12%.
Source: Tourism Research Australia
In addition to the data issues, there is also the ever-present demand for more granular and timely reporting of domestic tourism estimates.
The mobile positioning data has been developed in a partnership between the Australian Government (Tourism Research Australia) and DSpark, an Australian company providing data science consulting services. This new model for domestic tourism estimates has been in development since 2021 and was finalised during 2024. The first release of data will occur in June 2025.
The new model increases the frequency of reporting, reduces lag and improves accuracy and precision. It also reduces total costs of the project by 45%. Costs are reduced due to the smaller sample size in the surveys. Costs for labour/resourcing and sample attainment due to falling response rates have increased significantly for surveys in recent years.
The contracts and key data initiatives are managed closely by Tourism Research Australia. Data is governed by all privacy and data regulations and stored and managed accordingly. No personal identifiers are use or stored during the process.
Access to DSpark data is limited to subscribing businesses. Data is accessed through an API using R code. Tourism Research Australia are contracted to use data internally, but they also share data once the original DSpark data has been altered, for example by treating the data or using it as input to a broader model. In this case the DSpark data is being used as an input to a modelling process alongside survey data. For other customers, DSpark also offers dashboard style presentations for results, this is relied upon by those organisations that do not possess the required coding skills to use the API.
Data maintenance is ongoing and is conducted to manage any changes in customer characteristics of big data providers. This is monitored through stratified household survey collection ensuring representation across Australia. A major review of the project will be conducted every 4 years. All survey facets including sample attainment through to weighting and benchmarking are reviewed and monitored continuously. Survey questions and questionnaire design for the survey component are reviewed annually with input from all funding partners across the state and territory governments and Commonwealth agencies.
For the purposes of measuring household behaviours and characteristics, stratified street level household surveying has been the optimal approach. This approach ensures control of the sample attainment and face-to-face interviewing which is best practice. However, the costs of this approach are prohibitive, especially for large-scale surveys. This is why many collections in the past switched to telephone interviewing and the increasingly popular online approaches over recent years.
The advent of new administrative data sources provides the opportunity for a mixed mode collection with smaller survey sample attainment. This is supported by big data sources to provide a better and more sustainable cost-effective solution. From 2025 onwards, Tourism Research Australia’s modelled solution will include mobility data, along with stratified household surveying, census data and estimates of resident population data (between census).
The mobile positioning data included in the solution is obtained mobile network operator (data from cell phone towers), GPS and mobile Apps data and is combined with data from a stratified household survey. It is then benchmarked to resident population data from the census and estimated population.
The model uses telecommunications data from towers as the basis of the solution. This provides a consistent data series with large sample across the nation. While GPS and App data are typically inconsistent, these data sources do provide valuable insights into behaviour over time. For example, where mobile positioning data is limited in outback or remote locations, GPS and Apps data provides information to assist imputation.
The large sample of the dataset allows for benchmarking to be done at finer levels of detail. This further strengthens estimates. The mobile positioning data is weighted at the unit record level (each record assigned its own value) using resident population information by age and gender. This benchmarking is also undertaken at finer levels of geography; in this case, a statistical geographic area called a Statistical Area 2 (SA2). SA2s are geographic areas that have been split into groups where the population is between 3 000 and 25 000 across the nation. Australia has 2 310 SA2s, which are designed to represent a community that interacts together socially and economically. This benchmark process is undertaken daily, another advantage of the larger sample. The process of daily weighting allows for the provision or reporting of estimates for a day or days in output estimates, and for tracking daily changes in populations. This process also considers daily changes to the customer base from the commercial mobile positioning data provider.
The data produces estimates for domestic overnight trips and nights and domestic daytrips. In the mobile positioning data, the home location is identified by reaching a 90-day stay as per the Tourism Research Australia domestic survey. To be an overnight tripper the device needs to travel greater than 40km from home with a stay of at least one night away. Routine travel for work and study are excluded. Daytrips are defined as travelling at least 25km from the home location, with a duration of at least 4 hours, excluding routine travel for work or study and overnight stays.
In testing, the data has proven to be less volatile, is available soon after the reference period, and can be used at much finer levels of geography when compared with traditional data sources. Data confrontation in developing the mobile positioning solution involved analysing and comparing mobility estimates across Australia with other established data sources including the National Visitor Survey, accommodation and booking data as well as census and other population information.
With this project, Tourism Research Australia combines the strengths of the mobility data with those of a stratified household survey designed for tourism. This model allows for the collection of key tourism characteristics and helps track the representativeness of the sample base in the mobility data which can be adjusted if required. The model also allows for flexibility in adding domestic tourism spend. This could occur once financial transactions data has been identified as being fit for purpose, having gone through data confrontation and analysis across a time series.
The project has provided confidence that Tourism Research Australia has found a suitable administrative data source to cover one of the two major axes for domestic tourism, movement – the other being spend.
Importantly, the project has enabled Tourism Research Australia to replicate the existing definitions for domestic travel by Australian residents. Figure 2 shows the monthly time series results for domestic overnight trips of Australians travelling in their home state or territory, with comparison of volume estimates across mobility and the National Visitor Survey data. This is typical of results across the data confrontation process which was undertaken during the final stages of the mobility project.
Most administrative or big data sources are constrained in measuring tourism due to the tourism definitions requiring aspects across distance and time. These can only be met by mobile positioning data, which has consistent contact with the device. Other data sources provide intermittent data points and therefore struggle to meet tourism definitions.
The mobile positioning data also provides many benefits over traditional approaches, these include timeliness (available 4 days after the reference period), a large sample across all areas of the nation therefore allowing for more granular and regular reporting of estimates, and reporting of events and data at smaller levels of geography enabled by the significantly higher sample size in the mobility data.
Source: Tourism Research Australia
However, there are also challenges in using and maintaining new administrative data sources. Well-designed traditional survey collections use random sampling allocations to enable better representation of all sectors of the population. Administrative data sources, however, are not structured in the same way. The mobility data – despite its significant sample size – is a census of a commercial customer base. It has not been designed to be representative of the broader population. In addition, a commercial customer base may also change over time, for example, in terms of demographics.
In anticipation of this, in 2014 Tourism Research Australia added questions in the National Visitor Survey on mobile network ownership to track travel propensity across the different customer bases. Despite a large sample in each mobile network ownership group, propensity varied greatly (Figure 3). For this reason, Tourism Research Australia’s use of mobility data will include some form of monitoring and, where necessary, adjustment for bias in the customer base.
Source: Tourism Research Australia
While the mobility data has many strengths, it does not have the ability to fully replace key items collected in traditional survey collections for tourism. Examples of this include purpose of visit, activities undertaken, satisfaction and spend.
The mobile positioning data can, however, be used extensively in its own right to conduct in-depth analysis. For example, a quick response can be achieved daily for monitoring of natural disasters and recovery. It has also been shown to be beneficial in analysing events, monitoring, and measuring traffic and transport and looking at changing population density. In this case study mobility data is being used to support reduced mixed mode surveys, strengthening estimates, increasing reporting frequency and granularity while reducing costs and lessening the burden on the public.
For further information please contact:
Rod Battye, Manager, Data Innovation and Partnerships, Tourism Research Australia, Rod.Battye@tra.gov.au
This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.
© OECD 2025
Attribution 4.0 International (CC BY 4.0)
This work is made available under the Creative Commons Attribution 4.0 International licence. By using this work, you accept to be bound by the terms of this licence (https://creativecommons.org/licenses/by/4.0/).
Attribution – you must cite the work.
Translations – you must cite the original work, identify changes to the original and add the following text: In the event of any discrepancy between the original work and the translation, only the text of original work should be considered valid.
Adaptations – you must cite the original work and add the following text: This is an adaptation of an original work by the OECD. The opinions expressed and arguments employed in this adaptation should not be reported as representing the official views of the OECD or of its Member countries.
Third-party material – the licence does not apply to third-party material in the work. If using such material, you are responsible for obtaining permission from the third party and for any claims of infringement.
You must not use the OECD logo, visual identity or cover image without express permission or suggest the OECD endorses your use of the work.
Any dispute arising under this licence shall be settled by arbitration in accordance with the Permanent Court of Arbitration (PCA) Arbitration Rules 2012. The seat of arbitration shall be Paris (France). The number of arbitrators shall be one.
This document was produced with the financial support of the European Union. The views expressed herein can in no way be taken to reflect the official opinion of the European Union.