Esto también se puede leer en español.

Leer en español

Don't show this again

Internet Services Leer en español

Twitter's experiments pay off as user numbers grow

More people are flocking to the site on a daily basis.

twitter-money.jpg

Twitter reported its second-quarter earnings on Friday before the market opened. 

James Martin/CNET

Twitter isn't afraid of change. It constantly tests new features and has rolled out a revamped desktop site and a Olympics partnership this month alone.

All the experimentation appears to be worth the effort.

On Friday, Twitter said 139 million people logged into the app every day in the second quarter. That's up from the 134 million daily active users the company had in the first quarter. Twitter stopped reporting monthly active users in April. 

The new results suggest Twitter is making progress in efforts to attract and retain users, something it's struggled with in the past. Despite its popularity with celebrities, public figures and politicians, including President Donald Trump, the social network is smaller than its competitors. By comparison, Facebook has 1.59 billion daily active users

Twitter sees some of its biggest traffic during live events, such as award shows, sport games and television shows, when people gather on the site to tweet in real time. Twitter is trying to lean into that popularity. On Thursday, the company said it inked a deal with NBCUniversal to broadcast content about the 2020 Olympic Games in Tokyo

Still, navigating the flood of tweets can be confusing and the site can be filled with toxic harassment. With the help of its users, Twitter has been testing new features, such as adding color to replies and a tool to hide replies to tweets, to address some of those problems.

Look for those efforts to continue. On a conference call with analysts on Friday, CEO Jack Dorsey pointed to the ways Twitter's been trying to improve what users encounter and said the company will keep working on further refinements.

"We want to add a lot more controls within the timeline around relevance," he said. "What's most important for us is agility," or the ability for the company to "quickly experiment" and learn what's working and what's not.

A key part of that, Dorsey said, is an increased reliance on machine learning to deliver more relevant content.

"We'll be building more technology to address the problems" around behavior that violates Twitter's rules, he said. "Health [of the platform] is going to be an ongoing initiative for us."
 

Now playing: Watch this: Twitter to start hiding tweets that violate policies,...
1:04

Twitter has also started labeling certain tweets that violate its rules but are important to the public interest. The practice, introduced in June, applies to politicians who are verified and have more than 100,000 followers. The company rolled the rule out as it balanced the calls to ban the president for hate speech and spreading misinformation with the complaints from Trump and other prominent Republicans that Twitter has censored conservative speech.

With the 2020 US elections approaching, Dorsey said that Twitter will work on a number of fronts: identifying forms of misinformation and manipulation, ensuring transparency around who's posting what and challenging suspicious behavior.

"The No. 1 priority within elections is making sure we're protecting the integrity of conversations," he said.

Twitter has been making other tweaks to its site. This month, the company rolled out a redesigned desktop site in an effort to make it easier to navigate. 

The social network reported $841 million in revenue, beating the $829.03 million that analysts surveyed by Thomson Reuters expected. Excluding both certain expenses and a significant tax benefit, the company earned 5 cents per share, compared with Wall Street's expectations of 19 cents per share.   

The company's stock jumped more than 6% to around $40.65 per share in premarket trading.

Originally published July 26 at 4:14 a.m. PT.
Update, 6:03 a.m.: Adds comments from Twitter's conference call with analysts.