Dare Your Friends to Discover New Things
Intro
In today's social data revolution, the problem is not creating or accessing the data, but organizing it in a way that makes it useful. Traditionally, the search paradigm that lay at the center of content discovery has revolved around accessing the data; taking advantage of existing pervasive social interactions and focusing on organizing and presenting that data is a relatively new concept. Content discovery on the mobile is especially broken, since the traditional paradigm of search is extremely inadequate for the mobile use case. To address this issue, we propose a mobile application "Dare Me" that does not follow traditional paradigms but instead leverages social interaction, in the form of "dares" between users, and the resultant social data created to discover and present content - in this case, things to do - to the user.
Traditional Content Discovery
The most basic, pervasive, and traditional form of content discovery is search. As a society we have become comfortable with search even though, as information consumers in the information age, we deserve better solutions that are in fact well within our technical reach. Search requires users to first figure out what they want, and then search for it. In the problem we are addressing, discovering things to do, search demands that users first determine what they want and then query a search engine, when in reality if the user already had some glimpse of what they were interested in, they might not be trying to discover new content anyways! Even worse is search on the mobile phone, which combines all the problems that we have for search and throws in unintuitive UI and difficulty of use.
How Social Data Will Revolutionize Content Discovery
Enter the new paradigm for content discovery: recommendation. Search is clearly outdated for at least some purposes, and of late, recommendation has played a worthy substitute. There are three types of recommendations: algorithmic recommendations, reviews/ratings based recommendations, and social recommendations. Content discovery is generally moving toward recommendation, and in the future, perhaps in desktop search and particularly in mobile discovery, it may be the case that content discovery does not even involve search at all. Rather, a user's internet activity and his or her interactions with other users in his or her social graph will provide data that can be used to recommend the user anything he or she desires. The ideal (in regards to how far recommendation can go) would be that the user receives recommendations for the right thing at the right time every time, and never even has to search for anything themselves. However, as a society we are only beginning to explore these opportunities. Amazon and Yelp are examples of strong business models built around the first two types of recommendations. On the other hand, social recommendation is potentially an extremely effective paradigm that has not yet been explored in any great detail. We believe that leveraging interactions in a social graph can provide data that can be used to make effective recommendations. Since mobile discovery as a use case is particularly unsuited to traditional content discovery, yet provides even more social data, such as real-time geolocation data, and thus is even more conducive to recommendations, we chose to tackle mobile discovery with social recommendation as the central discovery paradigm.
The Application
When it comes to discovering things to do while on a mobile, options are extremely limited. Navigating straight to a website like Yelp to consider your options - bad. Using a search engine and navigating through pages that were never meant to be viewed on a mobile - worse. With Dare Me on Android, we sought to use social recommendation to create a mobile system of content discovery where the user does not have search, navigate web pages, or deal with any of the machinery that makes traditional content discovery on the mobile ineffective. The basic communication medium of our application is a "dare." If somebody sees a movie at the cinema and finds it to be exhilarating, he or she dares their friends to watch it as well. If you're out on the town and just finished a great meal, you dare your friends to check out the restaurant next time they're near by. The dare represents the central social interaction in our application, and we added features that encourage even more social interaction - and therefore more data that may be used for recommendations - such as forwarding and upvoting dares. Once a dare is received, a user may choose to accept the dare and complete the dare later using geolocation check-in or reject the dare. While we expect some dares to be rejected or ignored, we're optimistic that the convenience of the application combined with the knowledge that dares come highly recommended from people in a user's social graph will compel users to consider and complete reasonable dares. Ultimately, we hope that for mobile users seeking to discover new things to do, Dare Me saves them from having to attempt to use search by offering a solution that provides recommendations that are even more effective and interesting.
The Data
We had a couple questions that we wanted to answer during the start of this project. Would people be compelled by the premise of our application? And If so, how would people use our application? Through the pursuit of these goals, we gathered quite a few interesting insights. Our initial hypothesis was that giving someone a dare is highly psychologically a very strong incentive to complete the task, and it was what we wanted to build our application around. For the most part, this question remains unanswered. We would have to compare data from other social recommendation engines with ours. Without strong controls on the variables, it would be hard to draw any strong conclusions. We believe that the best way to answer our initial questions for the project would be to release the application and let the market answer them for us. Ultimately, those are the questions that any company has to answer when working on a product.
However, through the data that we did gather through our initial users, we found a few things that surprised us.
Popularity Draws Attention, But Not Action
We found that dares that were voted up more were often clicked through and viewed, but not accepted or completed. This conclusion was supported both quantitatively and qualitatively. Our numbers that we collected showed that votes correlated well with impressions on a dare, and there did not seem to be a correlation between votes and acceptance rates. From a qualitative standpoint, we spoke with our users as they used the app, and the general consensus seems to be that a dare has to be directly appealing for someone to accept it, regardless of its perceived popularity.
Popularity Is Less Important Than Source
For both views and dares acceptance rates, we found that a big factor in a user's decision to click-through or accept a dare was their relationship with the person that created the dare. Dares viewed from the "Friend's Dares" section had a higher click-through rate as well as a higher acceptance and forwarding rate. This fit well into our original hypothesis that social recommendations would be a more powerful and compelling tool than review/rating based recommendations
People Are As Interested In The Public As They Are In Their Friends
This result surprised us. We figured that from our above hypothesis, because people like recommendations from friends better than strangers, that the defacto section for users to go to would be "Friend's Dares". However we found that users were always exploring both public dares and friends dares each time they fired up the application. We hypothesize that this is partly due to how users consume mobile content, and partly the "Twitter effect", where users are used to browsing through a public stream and gathering information on people that might not be their friends.
People Forward Public Dares More Than Dares From Friends
We found this to be a possibly statistically insignificant insight, but still a very interesting one. A good number of the dares being forwarded (70%) were public dares rather than friend dares. When we pressed one of our users for a possible reason, he told us that he felt a friend's dare was "owned" by him/her, and that forwarding the dare didn't seem right. However, we believe that "Re-daring" will be an important part of the application in the future. Once users are used to the flow of the application, they will become more comfortable with sharing friends dares.
Users Are Reluctant To Produce
Users enjoyed participating, but were initially hesitant to create, which is very much in the spirit of the 90/9/1 rule. We were aware of this issue, and I think it is definitely a challenge for any kind of application that relies on user-generated content. Our users were very happy to browse through, accept, forward, and complete dares for fun. But we had to seed them with general ideas and prod them to create the dares. Ultimately, we have to iterate on our process to make it simpler and easier for a user to create dares and share them.
Finally...
The lifecycle of a social data project never ends. We hope that you have found value in reading about our final project for the Social Data Revolution. We're excited to see how recommendation and social data will continue to revolutionize content discovery.
In today's social data revolution, the problem is not creating or accessing the data, but organizing it in a way that makes it useful. Traditionally, the search paradigm that lay at the center of content discovery has revolved around accessing the data; taking advantage of existing pervasive social interactions and focusing on organizing and presenting that data is a relatively new concept. Content discovery on the mobile is especially broken, since the traditional paradigm of search is extremely inadequate for the mobile use case. To address this issue, we propose a mobile application "Dare Me" that does not follow traditional paradigms but instead leverages social interaction, in the form of "dares" between users, and the resultant social data created to discover and present content - in this case, things to do - to the user.
Traditional Content Discovery
The most basic, pervasive, and traditional form of content discovery is search. As a society we have become comfortable with search even though, as information consumers in the information age, we deserve better solutions that are in fact well within our technical reach. Search requires users to first figure out what they want, and then search for it. In the problem we are addressing, discovering things to do, search demands that users first determine what they want and then query a search engine, when in reality if the user already had some glimpse of what they were interested in, they might not be trying to discover new content anyways! Even worse is search on the mobile phone, which combines all the problems that we have for search and throws in unintuitive UI and difficulty of use.
How Social Data Will Revolutionize Content Discovery
Enter the new paradigm for content discovery: recommendation. Search is clearly outdated for at least some purposes, and of late, recommendation has played a worthy substitute. There are three types of recommendations: algorithmic recommendations, reviews/ratings based recommendations, and social recommendations. Content discovery is generally moving toward recommendation, and in the future, perhaps in desktop search and particularly in mobile discovery, it may be the case that content discovery does not even involve search at all. Rather, a user's internet activity and his or her interactions with other users in his or her social graph will provide data that can be used to recommend the user anything he or she desires. The ideal (in regards to how far recommendation can go) would be that the user receives recommendations for the right thing at the right time every time, and never even has to search for anything themselves. However, as a society we are only beginning to explore these opportunities. Amazon and Yelp are examples of strong business models built around the first two types of recommendations. On the other hand, social recommendation is potentially an extremely effective paradigm that has not yet been explored in any great detail. We believe that leveraging interactions in a social graph can provide data that can be used to make effective recommendations. Since mobile discovery as a use case is particularly unsuited to traditional content discovery, yet provides even more social data, such as real-time geolocation data, and thus is even more conducive to recommendations, we chose to tackle mobile discovery with social recommendation as the central discovery paradigm.
The Application
When it comes to discovering things to do while on a mobile, options are extremely limited. Navigating straight to a website like Yelp to consider your options - bad. Using a search engine and navigating through pages that were never meant to be viewed on a mobile - worse. With Dare Me on Android, we sought to use social recommendation to create a mobile system of content discovery where the user does not have search, navigate web pages, or deal with any of the machinery that makes traditional content discovery on the mobile ineffective. The basic communication medium of our application is a "dare." If somebody sees a movie at the cinema and finds it to be exhilarating, he or she dares their friends to watch it as well. If you're out on the town and just finished a great meal, you dare your friends to check out the restaurant next time they're near by. The dare represents the central social interaction in our application, and we added features that encourage even more social interaction - and therefore more data that may be used for recommendations - such as forwarding and upvoting dares. Once a dare is received, a user may choose to accept the dare and complete the dare later using geolocation check-in or reject the dare. While we expect some dares to be rejected or ignored, we're optimistic that the convenience of the application combined with the knowledge that dares come highly recommended from people in a user's social graph will compel users to consider and complete reasonable dares. Ultimately, we hope that for mobile users seeking to discover new things to do, Dare Me saves them from having to attempt to use search by offering a solution that provides recommendations that are even more effective and interesting.
The Data
We had a couple questions that we wanted to answer during the start of this project. Would people be compelled by the premise of our application? And If so, how would people use our application? Through the pursuit of these goals, we gathered quite a few interesting insights. Our initial hypothesis was that giving someone a dare is highly psychologically a very strong incentive to complete the task, and it was what we wanted to build our application around. For the most part, this question remains unanswered. We would have to compare data from other social recommendation engines with ours. Without strong controls on the variables, it would be hard to draw any strong conclusions. We believe that the best way to answer our initial questions for the project would be to release the application and let the market answer them for us. Ultimately, those are the questions that any company has to answer when working on a product.
However, through the data that we did gather through our initial users, we found a few things that surprised us.
Popularity Draws Attention, But Not Action
We found that dares that were voted up more were often clicked through and viewed, but not accepted or completed. This conclusion was supported both quantitatively and qualitatively. Our numbers that we collected showed that votes correlated well with impressions on a dare, and there did not seem to be a correlation between votes and acceptance rates. From a qualitative standpoint, we spoke with our users as they used the app, and the general consensus seems to be that a dare has to be directly appealing for someone to accept it, regardless of its perceived popularity.
Popularity Is Less Important Than Source
For both views and dares acceptance rates, we found that a big factor in a user's decision to click-through or accept a dare was their relationship with the person that created the dare. Dares viewed from the "Friend's Dares" section had a higher click-through rate as well as a higher acceptance and forwarding rate. This fit well into our original hypothesis that social recommendations would be a more powerful and compelling tool than review/rating based recommendations
People Are As Interested In The Public As They Are In Their Friends
This result surprised us. We figured that from our above hypothesis, because people like recommendations from friends better than strangers, that the defacto section for users to go to would be "Friend's Dares". However we found that users were always exploring both public dares and friends dares each time they fired up the application. We hypothesize that this is partly due to how users consume mobile content, and partly the "Twitter effect", where users are used to browsing through a public stream and gathering information on people that might not be their friends.
People Forward Public Dares More Than Dares From Friends
We found this to be a possibly statistically insignificant insight, but still a very interesting one. A good number of the dares being forwarded (70%) were public dares rather than friend dares. When we pressed one of our users for a possible reason, he told us that he felt a friend's dare was "owned" by him/her, and that forwarding the dare didn't seem right. However, we believe that "Re-daring" will be an important part of the application in the future. Once users are used to the flow of the application, they will become more comfortable with sharing friends dares.
Users Are Reluctant To Produce
Users enjoyed participating, but were initially hesitant to create, which is very much in the spirit of the 90/9/1 rule. We were aware of this issue, and I think it is definitely a challenge for any kind of application that relies on user-generated content. Our users were very happy to browse through, accept, forward, and complete dares for fun. But we had to seed them with general ideas and prod them to create the dares. Ultimately, we have to iterate on our process to make it simpler and easier for a user to create dares and share them.
Finally...
The lifecycle of a social data project never ends. We hope that you have found value in reading about our final project for the Social Data Revolution. We're excited to see how recommendation and social data will continue to revolutionize content discovery.
