Nieman Foundation at Harvard
HOME
          
LATEST STORY
With in-article chat bots, BBC is experimenting with new ways to introduce readers to complex topics
ABOUT                    SUBSCRIBE
Feb. 5, 2018, 1:03 p.m.
Reporting & Production

This hyperlocal news site in San Francisco is reinventing itself with an automated local news wire

Hoodline is focused on the local stories that can be found by mining large data sets, whether from city governments or from private companies like Yelp and Zumper.

It was once a San Francisco-based news site that covered news on a granular, neighborhood-by-neighborhood level. Hoodline’s work always had a technological bent to it, and its staffers looked hard at data and documents to surface the kinds of hyperlocal stories that people living in SF neighborhoods might want to know about. Did a favorite bakery re-open elsewhere? What’s happening with that new bus lane down the block? Those types of stories were Hoodline’s bread and butter.

Now, Hoodline’s major offering is an automated news wire service focused on the local stories that can be found by mining large data sets, whether from city governments or from private companies like Yelp and Zumper. It handles the raw data and data analysis side of things: collecting available public data sets, forming partnerships with private companies to use their data, and narrowing trends and other local news topics that might be worth pursuing as stories down to the neighborhood level. Then it generates articles from this cleaned-up data using automation software (with help from Automated Insights, also used by the Associated Press), which are then available to news organizations interested in data-backed stories helpfully localized for the communities they cover.

“In our earlier days at Hoodline, we tried to take a tech-heavy approach to help our work scale more. We were combining all the neighborhood blogs in San Francisco, we built our own CMS, we mined public data, we had full-time editors, we had freelancers in all the neighborhoods,” Eric Eldon, Hoodline’s editor-in-chief, said. (It still maintains some news coverage on its own site.) Finding a business model that could support hiring enough editorial staff to keep up high-quality neighborhood-level coverage became a roadblock. “By late last year, we had done enough tests with basic data sets that we realized: We could actually publish a lot of stories at once, based on analysis of information available in these data sets, and then feed these stories out to partner news organizations. We think this could, potentially, do a lot for local publishers. And fundamentally, the stories we do are not competing with the kinds of stories local news would do themselves.”

The origin story of the new Hoodline can be traced back to the late summer of 2016, when another startup, Ripple News, a newcomer to the local news scene, acquired Hoodline and much of its editorial staff (Ripple.co simply redirects to Hoodline now). Early on, Ripple had gotten into some trouble when it automatically scraped stories from the websites of news outlets like DNAInfo and Gothamist that hadn’t given the startup permission to reprint their stories. Joining forces, the Hoodline-Ripple News group still hoped to make the dream of a financially sustainable local news business that offers up substantive, original local stories work, and work at scale.

Last year, Hoodline joined a Disney accelerator. During that time, the team connected with Disney’s Wendy McMahon, now the president of ABC’s owned television stations group, which includes TV stations in Chicago, Fresno, Houston, Los Angeles, New York, Philadelphia, Raleigh-Durham, and San Francisco, and began testing its automated news wire service with these stations. (McMahon wouldn’t discuss the financials of the partnership.)

“Hoodline is so tapped in at the hyperlocal level. They have access to and are aware of news stories bubbling up at the neighborhood level that, say, we at KGO in the San Francisco and Bay Area might not be,” McMahon said. “What we’ve done with them works well from a resource-sharing perspective. If you think about our newsrooms: we need to be local brands. We have reporters who do their jobs fantastically well. But think about markets like New York or L.A., and how many neighborhoods there are in each, and that what’s relevant to someone in Orange County might not be relevant to someone in Pasadena or in Burbank. We can work with Hoodline to make sure news served to someone in Pasadena is Pasadena-based.”

The ABC channels have been in the process of improving their news apps to allow users the ability to better customize their news streams to their locations, which is where Hoodline also comes in: “If you say, hey, give me all the content for Orange County and L.A., there’s a good chance you’ll see Hoodline-created stories on Orange County. We’re using the partnership to help make these personalized experiences more robust.”

Local Hoodline-produced stories touch everything from activities to do in the Bay Area over the weekend to examinations of empty storefronts in a particular neighborhood, to a peek at the cheapest available apartments in a specific neighborhood (by the way, in this neighborhood, $3,768 for a one-bed).

Disney’s ABC has been Hoodline’s biggest newswire partner so far, but it’s also testing the waters with other chains like Digital First Media and Hearst. (Hoodline’s own blog post about its product cites some improvements in clickthrough rates. Let’s wait and see.)

“There are so many stories news organizations could potentially do, that nobody can afford to, because it’s expensive and time-consuming to have that many people on the ground,” Eldon sad. “We’re starting with the simple stuff right now: New business openings, rental price trends — simple story types that we can produce using data sets that cover a lot of geographical places and then distribute to a lot of people. Over time, we’ll want to get more sophisticated with how we analyze the data we have.”

The ultimate sophistication of the local news stories will depend in part on how much data Hoodline is able to collect and clean; it’s in talks with half a dozen other companies about access to their data, according to Eldon. It’ll also depend on what Hoodline’s data scientists and editorial staff are able to glean from their subsequent analyses, which is the main bottleneck in the process of generating localized data news stories. Private companies like Yelp or Better Doctor, another SF-based company that provides healthcare directories and doctor reviews, might give Hoodline access to their data, but those data sets are structured for internal use.

“Take a story, for instance, comparing the typical house you could get in one neighborhood versus another. If we want to do that story, we could take Zillow data, or other real estate site data, and try to extract what we can, and not just do it for one house or a few particular houses, but run large-scale statistics on the whole set,” Shwetank Kumar, Hoodline CTO, said. Work on its data wire service started about a year ago, and the company now has around 13 people on the tech team, including four data scientists.

Kumar and Eldon were quick to rebut a fear I hadn’t actually raised. It seemed they were anticipating some backlash to the idea of automating some parts of local news. They stressed that, in the case of Hoodline, robots are definitely not going to take over human journalists’ work. Hoodline’s local news wire is meant to amplify the work local reporters are uniquely positioned to do.

“We’ve gone through a phase to figure out the stories we could write that would be of local interest in the first place,” Kumar said. “We have to make sure we’ve tested that the stories we write are interesting enough.”

The data wire is still in its pilot stage, so the business model has yet to really be tested. It’s run affiliate links on real estate–related stories, for instance, pointing back to real estate listings.

“We’ve been playing around with the business model, either giving it away for free, or at an economical rate that would work for everyone,” Eldon said. “We think we can do this at a large enough scale that local publishers would be able to greatly increase the number of data-driven stories they’re doing within the geographical area they cover. It would allow them to cover more ground, and could help with their traffic, and all of that could help with their own existing revenue models, as well.”

Image of a skateboarder in the Lower Haight, by Brandon Doran, used under a Creative Commons license.

POSTED     Feb. 5, 2018, 1:03 p.m.
SEE MORE ON Reporting & Production
SHARE THIS STORY
   
 
Join the 45,000 who get the freshest future-of-journalism news in our daily email.
With in-article chat bots, BBC is experimenting with new ways to introduce readers to complex topics
“For us, this is a way to let people read and ask questions at their own pace, instead of having them read through long screens of text. Often people aren’t engaged in stories because they haven’t had the right context.”
Can we keep media literacy from becoming a partisan concept like fact checking?
Plus: Screen time debates, and what the data says about kids and smartphones.
After years of testing, The Wall Street Journal has built a paywall that bends to the individual reader
Non-subscribers visiting WSJ.com now get a score, based on dozens of signals, that indicates how likely they’ll be to subscribe. The paywall tightens or loosens accordingly: “The content you see is the output of the paywall, rather than an input.”