Forecasting city arrivals with Google Analytics

An article entitled “Forecast City Arrivals with Google Analytics” co-authored by Dr. Ulrich Gunter and Dr. Irem Önder, both Assistant Professors at MU's Department of Tourism and Service Management, was recently accepted for publication in Annals of Tourism Research. This journal is widely recognized as one the leading, if not the leading, academic journal in the tourism discipline worldwide.

There has been dramatic growth in Internet usage, especially for travel information search purposes. Each time an individual uses a website he or she leaves traces on that site. These traces can be collected and used for different purposes such as tracking user behavior, recommending products to the customer on their next visit to the website and optimizing website usability. Google Analytics accounts make it possible for businesses to collect these traces from their websites. Although many destination management organizations (DMOs) are collecting these types of information from their websites, this information is usually not used for making managerial decisions, but merely by IT departments to enhance website usability. Often, the interpretation of website traffic indicators such as Google Analytics is not clear to DMO managers. However, website traffic data can be very informative: showing, for instance, from which countries users originate. This information can then be combined to see if there is a correlation between the country of origin of website visitors and the country of origin of the actual arrivals to the destination. A primary objective of this article is therefore to show how website traffic data can be used by DMO managers to forecast tourism demand.

The theory explaining the search behavior of the visitors to the DMOs’ websites is called information foraging theory, which is derived from behavioral ecology and is similar to food foraging theories in anthropology. It states that, when possible, the information search process evolves toward maximization of relevant information per unit cost. In this respect, information foraged from a DMO’s website comes at a comparably low unit cost. In a tourism demand forecasting context, the information foraged from the DMOs’ websites and expressed by various Google Analytics website traffic indicators can therefore constitute an important predictor for actual arrivals to a destination if the foraged information proves relevant. Relevance of this foraged information, in turn, is given if considering it in appropriate forecast models results in comparatively more accurate forecasts.

Google Analytics is a free service by Google to the owner of a website that provides website traffic data including all user behavior on the website, such as the number of visitors, the time spent on the website, the number of actions taken on the site, where the users come from, etc. Google Analytics sends the website traffic data to the analytics server by means of a snippet (tracking code) that is included on the website and activated when a visitor views a page on somebody’s website. Overall, Google Analytics has a nearly 83% share of the website tracking tools market, yet this service is generally used only for website quality control and to enhance website user experience.

Website traffic data such as Google Analytics website traffic indicators can be used to develop forecasts which enhance managerial decisions. In this study, ten Google Analytics website traffic indicators (Average Session Duration, Average Time on Page, Bounce Rate, New Sessions, Page Views, Returning Visitors, Social Network Referrals, Total Sessions, Unique Page Views, and Users) were used to predict tourism demand in terms of total tourist arrivals to Vienna using forecast models able to deal with type of Big Data.

The results showed that novel forecast combination methods that include the Google Analytics information outperformed the univariate benchmark models for longer forecast horizons (three, six, and twelve months ahead), which are typically more difficult to predict. In addition, apart from almost immediate forecast horizons, forecast accuracy of tourism demand already benefits from the foraged information contained in the Google Analytics website traffic indicators for forecast horizons shorter than two years.

The use of big data in forecasting tourism demand is a new and important step for tourism research and the tourism industry. Since Google Analytics website traffic indicators were found to be useful in forecasting and the data are available as of today, nowcasting represents a new opportunity for the tourism industry. This study indicates that Google Analytics, or any other type of website traffic data, can not only be used by the IT department of DMOs such as the Vienna Tourist Board to enhance website performance, but also by forecasters to predict actual tourism demand to a destination. As a result of improved forecasting accuracy, tourism resources can be more efficiently allocated, cost reductions can be realized, and trends in tourist behavior can be detected earlier.