Twitter (TWTR) is trying to build a better experience for its users. The goal is to increase user engagement with the platform and grow the number of monthly active users (MAUs) and daily active users (DAUs). At the same time, the company is looking to drive its ad revenue higher by more effectively placing ads in user's Twitter feeds.
A couple of weeks ago, I wrote about Twitter's plans to increase live video broadcasts on the platform as a means to grow the user base, but this is not all Twitter is doing as it seeks to get people to spend more time on the platform.
Letting the machines learn
Back in February, Twitter management talked about plans to use artificial intelligence (AI) on its platform. In its fourth-quarter 2016 letter to shareholders, the company said it had seen an increase in number of users and time spent on the platform during the quarter and it attributed this good news to a better curation process to put relevant tweets in a user's timeline. "To build on this progress, we expect to apply machine learning more broadly across our service in 2017. Machine learning is critical for us to better identify and personalize content that people want to see and deliver it to them, faster."
Machine learning broadly refers to computers (machines) learning on their own and thus getting better at the services they provide. Twitter earlier this year also consolidated its AI working group under one manager, Jan Pedersen, who it hired from Microsoft in January. The AI efforts at Twitter have not gone totally unnoticed.
Catching the eye of billionaire Mark Cuban
On May 2, entrepreneur, investor, and TV personality Mark Cuban caught people by surprise when he told CNBC in an on-camera interview that he had recently been buying shares of Twitter: "I started buying Twitter recently because I think they finally got their act together with artificial intelligence." And maybe Cuban's right.
AI and live-streamed broadcasts could be two reasons the first quarter showed an acceleration in Twitter's MAU and DAU additions.
|Metric||Q1 2017||Q4 2016||Q3 2016||Q2 2016||Q1 2016|
|DAU growth (YOY)||14%||11%||7%||5%||3%|
|MAUs||328 million||319 million||317 million||313 million||310 million|
A group within Twitter that works on machine learning and AI is aptly named Cortex. In June of last year, Twitter bought London AI start-up Magic Pony Technology for $150 million. Previously, the company had purchased Madbits in 2014 and Whetlab in 2015 and formed a group to work on AI.
In announcing the Magic Pony acquisition, Twitter CEO Jack Dorsey wrote that its technology would help Twitter develop machine learning capability around videos and imagery. "Magic Pony’s technology -- based on research by the team to create algorithms that can understand the features of imagery -- will be used to enhance our strength in live and video and opens up a whole lot of exciting creative possibilities for Twitter," Dorsey wrote.
On May 9, Twitter engineers used a blog post to give a look at how AI is being used to show users more relevant tweets. Prior to the implementation of AI, a user's timeline showed tweets from people who were followed by the user in reverse chronological order. Using AI, Twitter now ranks every tweet based on a relevance model that predicts how interesting a tweet will be for a specific user. The tweet can contain video as well as text. The highest-scoring tweets are shown at the top of the user's timeline.
Twitter takes into account things such as how recent the tweet is, how often it's been retweeted, how strong your connection is with the tweeter, and tweets you've found engaging in the past. This helps the user see the best tweets first and the result to date is higher user engagement.
Now that Twitter has developed a capability to capture more of a user's attention, it is only natural that it migrates the technology over to its advertisements. Serving up the most interesting and relevant ads for a specific user should go a long way toward generating higher returns for advertisers. Management has already taken steps to help advertisers measure an ad's effectiveness.
Once advertisers can see improving results, it's not too much of a leap to see higher ad revenue for Twitter, which could spell good news for investors.