Travel and tourism forecasts become more accurate with analytics

DER Touristik builds web-based planning tool with SAS to better predict future demand and quickly react to changes in the market

Good planning is crucial in the business world — especially in volatile industries like tourism. Estimating the number of guests, revenues and margins in advance is complex and time-consuming if you do it manually. There is always a degree of uncertainty.

In recent years, DER Touristik Group has become one of the leading travel companies in Europe. It needed forecasting capabilities that could keep pace with its growth. The goal was a customized, web-based planning tool to simplify processes and provide accurate forecasts of key variables. The project started at DER Touristik Germany.

Traditionally, DER Touristik used spreadsheets for planning. Forecasting was based on simple mathematical predictions with empirical values. In practice, spreadsheet data that was three or four weeks old formed the basis for these rudimentary projections. Large discrepancies between forecasts and actual numbers were the result. To enable more reliable planning, DER Touristik wanted to consolidate this process and establish a single tool to handle all its forecasting needs.

Establishing an agile, automated forecasting system

“Data, tools, expertise – we had all of that in-house already. What we needed was a partner to develop the web application,” says Herbert Großmann, Project Manager for BI and Analytics at DER Touristik.

The decision to bring SAS in was an easy one, as DER Touristik has worked with SAS for decades, having built its entire business intelligence infrastructure on SAS. DER Touristik also was an early adopter of SAS Visual Analytics, and it now relies on SAS Viya, an analytic and data management platform running on a modern, scalable architecture. The historical data and current figures that the web application draws upon originate from the data warehouse of a SAS grid environment.

DER Touristik brought on Boemska, a global technology company recently acquired by SAS, which designed an HTML5-based planning tool and workflow with integrated forecasting that communicates with the SAS environment. Controlling at DER Touristik specified the functional requirements and was responsible for testing and approval. The Business Intelligence Competency Center (BICC) developed the ETL processes for providing the database and delivered the analytic component in the form of SAS Visual Forecasting on SAS Viya.

Agility was a crucial success factor, as additional requirements that were not included in the original specifications became known after work on the project had begun.

“We had a bunch of people from very different backgrounds on the team, but having regular meetings and processes to coordinate everything for the project worked out great,” Großmann says.

A preliminary run demonstrates power of analytics

To determine whether an analytics solution could improve its planning, DER Touristik ran the forecasts for the previous year again and compared the projections coming from the analytics platform with the original calculations. The result – SAS Visual Forecasting immediately delivered more accurate forecasts, deviating a mere 3% from actual bookings. Forecasts obtained using conventional methods missed the mark by 20% to 25% after adjustment.

“It was a real eye-opener for the planning department,” Großmann says. Next, it was on to developing and implementing the web application, which went fairly quickly. In under a year, DER Touristik saw the first results from its new system. Now many processes ranging from data integration to updating management are streamlined and automated.

Key figures at a glance for planning

The most important metrics for tourism planning are number of guests, revenue and margin. Forecasts are created based on historical booking data from the previous three years. Graphs of the most important metrics let planners see, at a glance, the impact of the values they enter into the system.

Thanks to analytics, DER Touristik has greatly simplified its planning process. Now forecasts are fast and precise. The tool also functions as a single source of truth for planning, providing the most current data. The quality of the data has also risen, as it is input directly with no aggregation necessary.

Nearly 80 users are now on the tool, most of them working inside departments or on the management team. Any SAS user can access the data from the interface. The application lets administrators determine who has which permissions to view or edit which kind of information. 

Covering industry requirements 

One special topic that tourism planners need to consider are school breaks. The home markets for DER Touristik are Germany, Austria and Switzerland. These countries vary greatly with regard to school vacation dates. In addition, the impact on bookings depends on which federal state in Germany is going on break. North Rhine-Westphalia, the most populous German state, has a much bigger effect than Bremen, for instance. The statistical representation of these effects that factor into the forecasting became a metric dubbed the Vacation Index.

According to Großmann, one of the biggest benefits of the planning tool is its ability to automatically disaggregate forecasts. This process involves breaking down seasonal data into individual months, which is the only way to verify that forecasts are accurate. Before introducing the tool, controlling had to spend a great deal of time breaking the numbers down manually.

“We had one staff member doing this for two straight weeks every season,” Großmann said. 

Planning during uncertain times

Tourism is subject to unpredictable events, such as natural disasters, political instability and pandemics. “Forecasting models are only informative to a certain extent during a global crisis,” Großmann explains. “The most important thing is to have models you can modify quickly as events unfold.”

When it comes to unpredictability, one important aspect is achieving a higher forecasting frequency. This is now possible because the forecasting process takes much less time. Before using SAS Visual Forecasting, DER Touristik only generated forecasts three times a season. The pre-season planning served to set budgets. Then two additional forecasts during the season took existing planning data and used it to generate figures extending all the way to the end of the season. Now the data is enriched with values that SAS Visual Forecasting proposes once a month. This allows forecast figures to be adjusted on the fly for more reliable planning. “We can really stand behind what we’ve achieved here,” Großmann says.