The Associated Press Leaps Forward

Case Study

AP now produces 3,700 earnings stories quarterly – a 12-fold increase over its manual efforts.

AP - Natural Language Generation The Associated Press describes itself as “one of the largest and most trusted sources of independent newsgathering, supplying a steady stream of news to its members, international subscribers and commercial customers.”

The winner of 51 Pulitzer Prizes, its members include about 1,400 daily US newspapers and thousands of television and radio broadcasters.

The company is known for style as well as content. AP Style is the definitive writing standard for journalism and many other fields.
 

The Challenge

Every quarter, public companies in the US release corporate earnings. And every quarter, Associated Press reporters ploughed through those reports, extracting the relevant financial numbers to compose stories based on those numbers.

There were two problems with this approach. First, AP could only produce 300 such stories per quarter – that left thousands of potential company earnings stories unwritten.

Second, these stories were not only taking up reporters’ time – they were, in the words of USA Today’s Roger Yu, “the quarterly bane of the existence of many business reporters.”

New York Magazine’s Kevin Roose remembers corporate earnings recaps as “a miserable early-morning task that consisted of pulling numbers off a press release, copying them into a pre-written outline, affixing a headline, and publishing as quickly as possible so that traders would know whether to buy or sell.” On TechCrunch, Alex Wilhelm notes that “[e]arnings season makes most reporters want to poke their eyes out with sharp objects.”

AP and its reporters needed to a find a better way. Fortunately, the company has a proud history of innovation. “One reason for AP’s longevity,” it says, “has been its ability to adapt quickly to new technologies.”
 

The Solution

AP found answers in automation. Specifically, the Wordsmith platform from Automated Insights.
Wordsmith uses natural language generation to turn data into a written, plain-language narrative. In this case, Wordsmith transforms raw earnings data from Zacks Investment Research into a publishable AP story in a fraction of a second. In fact, the Wordsmith team specifically configured the natural language generation engine to write in AP style.

As a result, AP now produces 3,700 quarterly earnings stories – a 12-fold increase over its manual efforts. The stories retain the same quality and accuracy that readers expect from any of AP’s human-written articles. Aside from an explanatory note at the bottom of the story, there is no evidence they were written by an algorithm.

After AP announced its “leap forward in quarterly earnings stories”, media outlets like The New York Times, Slate, and Mashable started writing about its innovative approach.

New York Magazine’s Roose called automated reporting “the best thing to happen to journalists in a long time.”

“So bring on the goddamn robots,” said TechCrunch’s Wilhelm.
 

The Results

In January 2015, Assistant Business Editor Philana Patterson explained automation’s numerous benefits.
Better for Customers
“There has been a great deal of interest about how automation works from both members and readers, and overall the reaction has been incredibly positive. AP members are getting more stories about companies in their markets than ever before.”
Better for Reporters
“Internally, the reaction has been positive from staff, largely because automation has freed up valuable reporting time and reduced the amount of data-processing type work they had been doing.”
Fewer Errors
“Once you set up automation, and go through a rigorous testing process, you reduce the prospect of errors. In fact, we have far fewer errors than we did when we were writing earnings reports manually.
Greater Efficiency
“We estimate the automation of earnings reports has freed up about 20 percent of the time that we had spread throughout the staff in producing earnings reports each quarter.”
Patterson also pointed to more opportunities in the future:
“This quarter, we are testing the automation of earnings from Canadian and European companies. We expect to add further enhancements and more companies in future quarters. My colleagues in the sports department are also exploring small-audience sports for automation in order to free staff to report news that fans and consumers do not get on the field or a broadcast.”
Automation hasn’t displaced any reporters, but AP says it has freed up the equivalent of 3 full-time employees across the organization. As The Verge noted, “computers are not taking journalists’ jobs — not yet, at any rate. Instead, they’re freeing up writers to think more critically about the bigger picture.”

“We couldn’t be happier,” AP vice president Lou Ferrara told The Washington Post. “When you look at the 170-year history of a company like the Associated Press, part of its success has been adapting and innovating in new media formats. . . We see this as part of that history.” 

Addendum: The Future of Media Automation

In late 2015, Automated Insights unveiled a new version of its Wordsmith platform, designed to make it easy for any professional to upload their own data and automatically generate their own narratives. The AP was one of the first companies to take a look.

In a working session with Ai, AP staff members, including the automation editor, explored how the new Wordsmith could generate stories using federal unemployment data. While AP reports on monthly unemployment rates for many states, it isn’t able to provide that coverage for all states. AP is testing out Wordsmith to see how it may automatically generate data-rich reports for all states each month.

Barry Bedlan, an AP Deputy Director and one of the organization’s proponents of automation, was impressed by Wordsmith’s ability to account for possible data scenarios.

“ Wordsmith allows you to make the if/then decisions that a reporter would make when writing a story,” he said. “By mapping out the editorial decisions in advance, it can be used to quickly cover most scenarios and then make any changes as needed. Ultimately, I can see how Wordsmith can be used for expanded coverage at an unprecedented speed and scale.”

The overall goal: for reporters to focus less on numbers and more on nuance, and for the AP to deliver more value to the news organizations that rely on them every day.

. . . 

For more on Ai”s relationship with AP, see our videos from Strata and SXSW.