Behavioral data part - results

In case you have missed the first two parts, you can find them here and here.

Introduction

In this part, we will present the hypotheses and the main insights from the survey and the behavioral dataset on the relationship between consumers' travel planning behavior and their personality.

Next to that, we will demonstrate and evaluate the performance of the model as well as where the findings can be applied in business.

Additionally, we will plot two respondents profiles – selected out of the data – who provide a good representation of the impact of risk and uncertainty on travel related online behavior. These profiles show their cross-device activity with the travel related online activities.

Sample characteristics and survey results

The sample of the research consists of Dutch panelists, equally distributed across gender and slightly more dominated by senior people with a median income between 39K and 66K Euro. 61% percent of the sample indicated they are the main decision makers concerning their household’s travel decisions.

The most visited destinations were in Europe accounting for 79% of the visits, followed by North America and Asia with 7% of visits each. Among all countries, Spain is the most popular destination with 20.5%. The domestic travel in the Netherlands accounted for 16% followed by Germany, Italy and France with approximately 10% each.

Based on the survey, the spread across the different segments of risk and uncertainty attitudes is as follows:

Main results

The main results from the analysis of the survey and behavioral data confirm most of the hypotheses that we made regarding the impact of personality in terms of risk and uncertainty on online travel related behavior.

Zooming into more details, the data has been analyzed from the perspective of desktop activity, mobile activity and both activities combined. The mobile dataset doesn’t contain enough observations and the variance is too low to make concrete conclusions about the mobile behavior of the participants. We have 426 participants among which we have mobile data for only 100 panelists. Therefore, the hypotheses are being confirmed based on the analysis of the desktop data. The most impressing insights are the following:

Based on the results we can confirm the following hypotheses:

According to our estimation based on the analysis we can compare consumers who made 3 bookings per year, are highly active online and differ in risk and uncertainty attitudes. Here is how their personality would impact the travel planning activities for one year:

Example profiles

In order to present to you how the risk and uncertainty impacts the online travel behavior we will introduce two real people from the data. The names have been changed to protect the persons' personalities.

Meet Marilou. She is a 34 year-old woman from Limburg. She earns above average income and she is highly active online in comparison to her peers. Marilou is an open-minded person with an outgoing personality. She is a social person with a lot of friends. She is curious and open to new experiences, however she holds a strong opinion about her beliefs and she tends to be suspicious. Marilou is both averse towards risk and uncertainty. She spends a big proportion of her time deciding on her travel activities, including searching for and comparing hotels. Apart from this, she also uses travel magazines in order to support her decisions. Marilou went on a leisure trip to the US with her significant other. She spent more than half a year on the planning her 18 nights stay even though she has been there twice before. She ordered everything online including flights, transportation, accommodations and entertainment. They spent approximately 3,000 EUR per person on the whole trip including transport and hotels.

Here is Marilou’s activity on travel related websites as a proportion from her total online activity, both on mobile and desktop.

Meet Sophie. She is a 27 year-old woman from South Holland. She earns above average income and she is not very active on the internet compared to the rest of the sample, however, she uses a lot of different sources of information. Sophie is very kind and trustworthy, perhaps a bit naive, sometimes a bit spontaneous, yet she is a stable individual. Sophie is averse towards risk, but uncertainty seeking. Although she was the main decision maker regarding her trip and she uses only internet for planning, she didn’t spend that big of a proportion of her time to decide on her travel activities comparing to Marilou. Sophie went on a short trip to the US with one more person. She spent only a week to plan her trip. She ordered accommodations and entertainment online. They spent approximately 900 EUR per person on the whole trip including transportation and accommodations.

Here is Sophie’s activity on travel related websites as a proportion from her total online activity, both on mobile and desktop.

The two graphs clearly indicate the impact on the travel related behavior and the difference between uncertainty averse and seeking attitude. They match with the finding that in terms of micro-moments there is no significant difference between both constructs. In terms of time and pageviews, the aversion plays a big role and increased them significantly comparing Marilou and Sophie.

Furthermore, one can see there are similarities between desktop and mobile behavior, perhaps slightly lagged. Also it is obvious that the graphs of Marilou and Sophie have several common peaks at weeks 1, 6, 14, 20 and 26. We can speculate that this activity is related to promotional activities from a travel service provider happening at the same time.

Conclusion and recommendation

Behavioral data gives us an amazing capacity to understand the online behavior of the consumers as never before. This research confirms the finding from Quintal (2010) ¹ in the area that the attitude towards uncertainty, but not risk is responsible for choosing a number of sources (in our case in terms of domains) used for travel decision making. Our findings further contribute to the area through passive metering by investigating the time people spend on travel websites, as well as the efforts in terms of pageviews. The results from the analysis strongly confirm our hypotheses that both risk and uncertainty attitudes account for changes in time and pageviews spent desktop devices.

The findings are applicable to online travel stakeholders in order to better target the consumer. For example, personalized advertising is an approach widely used by the big travel websites. The insights revealed by this research can help businesses craft better personalized communication and target the right content in the right time to the right consumer.
Based on the behavioral data, a service provider can estimate the profile of the consumer in terms of risk and uncertainty attitudes. Once this profiling information is available, consumers can be targeted with a message containing appealing offers based on their profile. Furthermore, the message can be delivered within a time interval when the probability to book is the highest, which could potentially result in better conversion.

In conclusion, we can determine from the analysis that mobile and desktop behavior differ at large and they have to be combined in order to produce meaningful insights. Yet, due to the difference in the behavior over both types of devices, there are important differences that have to be acknowledged. For example, we were able to detect bookings for each participant only over desktop device and only one purchase over mobile. In this case study the hypotheses were confirmed over the desktop activity. As a recommendation for future research we recommend more extensive research over mobile devices as we had only 25% of the sample using both mobile and desktop.

Limitations

An important limitation of this research that has to be acknowledged is the fact that risk and uncertainty have been assessed on scales used to measure “General risk and uncertainty attitudes” as we aimed to keep the questionnaire as short as possible. Decisions under risk and uncertainty is a big branch of science that offers advanced measurement techniques, such as lotteries and bets. Furthermore, there is evidence that domain specific risk attitude can differ from general risk attitude, therefore measuring perceived risk and uncertainty towards travel can also add up to this research.

You can download the handout of our case study here.

References

The post has been originally published as a marketing copy at Wakoopa's blog How behavioral data sheds light on real consumers' travel planning behavior – Part III on 16 September 2016