(Bloomberg) -- Xavier Grangier, the chief technology officer and head of digital at Liberation, the venerable French newspaper, had a problem. Like many publications, Liberation was struggling to adapt to a world in which readers were increasingly finding content via social media.

"We had RSS feeds, but we weren’t happy with that," Grangier said in an interview. Liberation also had a team that posted a selection of articles, chosen by the paper’s editors, each day to social media, but the process, he said, was a hassle to coordinate.

In January 2015, Grangier turned to Echobox, a London-based startup that uses artificial intelligence to manage the publication of editorial content over social media. The software determines the most opportune time to post a particular story to drive readership, can recommend what headline or tweet to send out, and can select the best photograph to illustrate the post.

Using the software to post an average of 27 articles per day, Grainger said that Liberation had seen a 37 percent increase in the number of people it reached on Facebook and a 42 percent boost in its reach on Twitter.

"We have way more articles being seen by 100,000 people or more than before," Grangier said. He also said it made life easier for his digital editors, allowing them to spend more time curating the stories they wanted to publish to social media and less on the logistics of actually posting that content.

Just when some traditional publishers were beginning to find their footing online, with large audience figures and pay walls, the growing influence of social media has threatened to crush this embryonic stability. Echobox thinks it can use artificial intelligence to help publishers regain their momentum.

Antoine Amann, Echobox’s founder, said he got the idea for the company in 2013 while working briefly as a reporter for the Financial Times, seeing how editors were wrestling to manage social media feeds. To build the artificial intelligence software that drives Echobox, he teamed up with University of Cambridge computer science PhD Marc Fletcher the following year. Fletcher is now Echobox’s chief technology officer, while Zoubin Ghahramani, a professor with Cambridge’s machine learning group, serves as a technical adviser to the company.

"We had to build a system that understands content like an editor could," Fletcher said.

Technical Core

The technology caught the attention of Michael Jackson, a partner at Luxembourg-based venture capital firm Mangrove Capital, which in July lead a $3.4 million financing round for Echobox. Amann had previously received some initial seed capital while incubating his idea through London-based Entrepreneur First.

"This is not just a sales team," Jackson said. "They have a core of developers who know what they are doing at a deep technical level."

Competing companies -- such as SocialFlow and Buffer -- provide analytic tools and suggest optimal times to post content. Some even provide forecasts of how well a particular piece of content will do. But Echobox professes to offer a fuller range of automation than those services, with its software able to alter a posting schedule to adjust to breaking news, posting content related to that event and delaying publication of less relevant stories.

Echobox uses a neural network, a type of machine learning that is designed to mimic the way parts of the human brain works. This system first learns the audience composition and reading habits for each publication and then makes predictions about the best way to optimize a particular story for social media. Over time, the predictions should get more accurate as it "learns" the nuances of the brand’s audience.

Echobox has about 100 customers so far, ranging from French newspaper Le Monde and La Nacion in Argentina, to the The Straits Times in Singapore. Some, such as Le Monde, allow Echobox to read its entire news feed and decide when and how to post articles to social media. Others, such as German science television show Galileo, use the software primarily to test which alternative headlines or photo illustrations will best drive traffic for stories that editors have already selected to publish.

Echobox charges customers a subscription fee based on the number of page views they receive, with prices increasing as traffic climbs. The company refused to offer specifics on how much it charges.

Johann Bayerl, the head of digital at Galileo, said in an interview that the program’s website gets about 80 percent of its traffic from social media, so maximizing that impact was critical. "We want to test audiences and headlines and pictures," he said, adding that Echobox was the only social media automation software he had seen that allowed this sort of A/B testing. He said Galileo saw a 20 percent improvement in its Facebook engagement in the first few months of using Echobox.

But Bayerl also pointed out some limitations of the system. After Facebook changed the algorithm it uses to manage users news feeds in June, Galileo saw a temporary drop in traffic, even using Echobox. Amann said that because Galileo is different from many of its other customers -- a television program instead of a newspaper, German language instead of English -- it took Echobox’s neural network longer to learn how to adapt to the Facebook algorithm change for Galileo.

"The vision for Echobox is to make sure that journalists and editors don’t have to think about things like an algorithm change at Facebook anymore," Amann said.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access