Twitter is a great Newsstand??
This a guest post, written by Henry Vuong from one of the teams on the Social Media Monetization case during Fall 2017. Lightly edited by Rohan Trivedi and Karthik Bala.
Overview
From 2016 to 2017, Twitter’s revenue declined significantly, and they have failed to acquire more active users. With lagging user engagement, what added features will help Twitter develop into a more engaging platform? Twitter’s “newsstand” stands out as a feature with potential to be improved. In this post, we will outline our business plan and Java implementation of a personalized newsstand.
Market Research
A study on social media platforms as a news source by Pew Research Center found that nearly six out of ten (59%) of Twitter users relied on the site for daily news (Facebook is at 66%). This puts Twitter in the top three social media sites according to percentage of users who rely on the site for news (Reddit is at 70%). From this data, it’s evident that Twitter’s best bet for recapturing users is through focus on news distribution.
Our research focuses on Twitter’s mobile app as it is responsible for the majority of Twitter’s ad revenue. The current news/explore tab showcases many characteristics of an efficient news source. Top stories are displayed prominently for the user, and there’s a news aggregator labeled “Popular articles”, which aims to keep individuals up to date on popular news, tabloids, or stories. However, lack of user personalization makes the news/explore tab less engaging. An alternative option would be to ask the user for topics of interest. Unfortunately, this will not create any competitive advantage for Twitter compared to like AP Mobile and Google News/Reader.
Development Process
Our solution is add personalization to the Twitter newsstand, tailoring displayed articles to users preferences. We accomplish this with a news algorithm that leverages user click data on articles, retweets, likes, accounts being followed, etc to determine which article types to display in the top news stories. Favorites and retweets would be weighted heavier than likes and followed accounts.
The news algorithm would then reorganize the top stories and be the new effective display. Our development team implemented a rudimentary version of the algorithm in Java.
One issue we encountered was how would we deal with news identity. A single article can talk about politics and technology which would make categorization, and thus personalization, difficult. The solution was to give articles the opportunity to display more than one news metatag.
Conclusion
The project was largely a proof of concept and there are obviously improvements/additions to do. One possible improvement would be to allow visualization of user inputs. Allowing users to see the number and type of articles they have been interacting with would give users positive feedback for interacting with articles, and users may favor Twitter’s newsstand over others simply because Twitter tracks your reading progress (and maybe even gives achievement awards).
Here is a link to our product: monetize-twitter
