finding target users through mixing data and user studies
Project Overview
In 2014, Airbnb was going through the crucial stage of transitioning its user base from the group of innovators to early adopters in Asia. In the book ‘Crossing the Chasm’ by Geoffrey Moore, this transitional stage is described as the most critical stage in the technology adoption lifecycle. To make a successful transition, it's important to understand who you must target, and hone in the effort to win over the target segment.
I was a Business analyst responsible for producing quantitative user analysis. I worked cross functionally with the research, operation, marketing, growth, and the PR teams. My user insights report was leveraged by the above mentioned functions to make strategic decisions for key countries during the Asia trip.
Leveraging the Word of Mouth effect
I was initially approached by Patrick Lee (Korea country manager) to help identify the right target group in Korea. Patrick strongly believed in the diffusion of innovations theory by Geoffrey Moore and understood that a successful international expansion must start by identifying the key group of people to target.
There were three main reasons why Patrick believed targeting a specific group of users was critical to Airbnb's success in Korea. First, targeting a small group of people will maximize the potential word of mouth effect. Second, a winning over a right group of users will lead to adoption by other groups. Third, there wasn't enough budget to target everyone. I was tasked to leverage the big data to pinpoint the users in Korea.
Understanding users through interaction
In order to accurately analyze users through using the data, it is critical to understand the users through personal interactions. This is why one of the first thing that I did for the project was attending Airbnb events that were held in Seoul, and talked to the users. In addition to attending events, I helped user researchers to execute usability tests which helped me to gain further insights to what the current users felt about Airbnb.
Understanding users through data
After realizing that not a lot of existing users have booked a trip through Airbnb, I looked at the internal data to see the distribution of the number of travels by our users in Korea. Given that the end goal of 'finding the target users' is to make them talk positively about Airbnb, it was important that the target users already used the service. Afterall, it is much easier to make someone talk positively about the product, after he/she tried the service. The difficult part is getting someone to try the product in the first place.
As I bucketed the users by the number of travels they made, I realized the importance of the 'first time' bucket. I found the first time bucket particularly interesting because those were the users who tried booking through Airbnb for the very first time. In other words, the behaviors by the first time users is a great indication of how the market thought of the service in the first place. To further understand how the market sees the product, I bucketed the first time travelers by different age groups, and dissected their user behavior.
As I cut the first time travelers by age groups, it was immediately apparent that one age group took a huge chunk of the overall first time booking. Once I was able to identify that a specific age group was much more likely to book, I analyzed how the gender was split for that specific group. From this analysis I found that for the key age group, one gender dominated, and was able to narrow down our target group by age, and gender. With the information about the target group's age and gender, I looked to the behavioral pattern for the group, such as booking price, # of nights booked, destination, and their Facebook likes.
From the behavioral analysis, I was able to narrow down the target user group to something like: 'Men in the age of 35-44 who liked wine, yoga, pilates, who travel mostly to Brazil as their first destination using Airbnb, and rent out a $300 listing per night and stay for 20 days'. (for confidentiality reasons, this particular example differs significantly from the actual user profile).
Combining disparate informatioN
While it was possible to narrow down the potential target group through demographics and behavioral analysis of the first time users, it was still too broad for us to target. After all, there's a lot of different type of users who fit into the bucket of 'men in the age of 35-44 who liked wine, yoga, pilates...etc'. For instance, both an investment banker, and a chef fit into those description. With the limited budget, trying to win over the entire group that fit the example, would be inefficient.
To go around this problem, I analyzed the top 50 traveler in the country through combining the information on the profile page, Linkedin, and Facebook information. I was especially intrigued by these users because they probably knew the product the most, and would be the best starting point to grow the users. Combining all these information produced a powerful understanding of the types of super travelers in Korea. Surprisingly, these 50 users fit into just a handful of different types of users.
Once I successfully narrowed down the 50 top travelers to four different user groups, I analyzed their behavior on the site, and compared the result to the average travelers in Korea. It turns out there was a significant difference in how different user groups were using Airbnb. For example, an english teacher (foreigner), may pay 40% less than the average price and travel only to Seoul. Through reading the chat history, and combining with the fact that majority of bookings happened over the weekend - I was able to understand how Airbnb was perceived by them.
Closing in on a target
Once our top users were segmented into few buckets, I was working with the Korea team in choosing the specific target group to focus the energy. There were multiple variables that were critical in forming the target from the handful of user groups. These were some of the main consideration when we chose the group:
First, we considered the possible negative implication our target group may have. What we didn't want was winning over one group, but losing out on another group because of the negative implication associated with the group we chose. For example, we didn't choose 'fashion models' as our target group because some people view 'fashion models' with a negative views.
Second, we chose the group that is in line with our brand position. We did not want to become a discount brand. Hence, it was really important that our target group spent above certain price point.
Third, we chose the user group that had significant social presence in the culture for our broader target group. As mentioned previously, through my initial data cut, we were able to hone down the characteristics of our first time travelers (e.g. 35-44 men who like wine, yoga, and pilates). We wanted our target group to have a significant social presence for that particular age group.
Through piecing together these variables, we were able to micro target a very specific set of people. For instance, using the examples above, our target group would be 35-44 designers who liked wine, yoga, and pilates. Because we had a specific group of users that we wanted to target, we could pinpoint our offline operational efforts to win over the users.
Fastest growing major country in the world
In 2H14, Korea was the fastest growing major country in the world at Airbnb. The impact of completely winning over a group of people was clearly evident, as each operation campaign resulted in 3X performance in comparison to a similar campaign in the next best performing country. I went on to produce similar type of analyses for all the key markets in the Asia Pacific region.