Today, we’re launching a data-driven local wire service for both media and tech companies.
Hoodline is turning nationwide data sets generated by tech products into stories that can help media organizations cover more ground, break news, build audiences, and even make money. (Check out TechCrunch's take, here.)
From business openings and trends derived through Yelp data, to neighborhood apartment prices and availability from Zumper, we’ve built a system in which our data science and editorial teams can produce thousands of automated, street-level stories across the US.
A group of new Japanese restaurants in Philadelphia that we identified by mining Yelp data, then published with ABC13.
By integrating with our location platform API, publishers can feed these stories into their content management systems, as they do with other wire services.
This allows media organizations to publish a higher volume of valuable local content to monetize via their existing revenue streams. In addition, some of these articles can include deep links or affiliate links related to the topic.
We’ve gotten strong early results from testing hundreds of articles with publishers around the country, including ABC/Disney, Digital First Media and Hearst, and on our own Bay Area neighborhood news site, and we’re beginning to work with more media companies such as McClatchy.
A Mercury News article looking at rental prices in a Silicon Valley suburb, based on Zumper data.
We’ve seen similar traffic performance compared to article averages overall. Engagement is notably high, with click-through rates on deep links averaging 15 percent—and some story templates seeing double that.
How The Wire Works
Our experiences covering San Francisco and Oakland neighborhoods have helped us understand the topics that interest readers about the world around them. To support that reporting, we’ve used public data sets to break stories for years. But while these data sets are quite valuable, they’re typically smaller, messier, and less accurate than the data we can get from tech products.
A front-page story in the San Francisco Chronicle last year, that used our analysis of Yelp data to examine trends in local coffee shops.
Around the country, tech products already provide local media with important insights—think of all the times you see data and research from Zillow, Redfin, Buildzoom, Yelp or other sites mentioned in local news coverage. But at the moment, these collaborations take a lot of work on both sides, and are typically one-offs. We can make this connection scale.
We’re trying to get access to as much local data as possible, so we can produce the best analyses. We’re ingesting terabytes of data per month from a range of sources: in addition to Yelp and Zumper, we’re also working with Groupon, Apartment List, ZipRecruiter, and more.
An internal Hoodline tool that visualizes the local news that our platform can distribute to any partner.
Once our engineers ingest the data into the platform, our data scientists and editorial team work together to produce large-scale data analysis across key local topics, organized down to the neighborhood and address level.
The location platform we announced last year is crucial to this effort. It lets us geofence data around local topics, based on locally preferred neighborhood boundaries and names. This allows us to provide granular coverage of trends and changes down to the neighborhood and address level. (It is also being adapted for other use cases, like showing relevant news articles in search products.)
Once we complete our data analysis, our editorial staff uses text templating software to create article templates that use the data to generate articles for distribution through the platform API.
Local publications can publish these stories as-is, as supplements to their existing coverage, or build on them with additional reporting.
We believe this approach will help address two key problems in local media. Anyone reading this knows that the challenging business model for local media has resulted in thousands of job cuts and publication closures in the last decade, with news deserts now spanning much of the country.
ABC30 in Fresno turned our story about a new business opening into a video interview with the establishment, and published as a combined package.
To take advantage of what’s possible online today, publishers need more local traffic to continue selling their local ads against, or to funnel towards subscription products. When it comes to getting reader payments, the information has to be fresh and relevant, and must reach enough local users to make the local funnel generate real revenue.
We think the unit economics of our stories, combined with the value of the information they contain, will help make these sorts of efforts possible.
We’re also excited about what we’ll be able to offer as we get access to more data sets.
A collaborative investigation we did with ABC7 Bay Area last year, using Yelp data combined with our own data and analysis to look at retail vacancy problems in San Francisco’s Castro District.
We have focused so far on the simplest types of stories, covering news so granular that it’s rarely, if ever, covered by others. But we’re also planning to combine data sets, to produce localized analysis in ways that are often only done in academia. Our vision is to expand data in local news beyond weather reports, election results and sports statistics, incorporating many more data sets both individually and in concert with one another.
Whether it’s us interpreting the data, or other tech and media companies using it as the basis of their reporting, we believe the final result is people getting a much better idea of what’s happening in their cities.
If you are a media or technology company interested in working with us, please email firstname.lastname@example.org.
For more details on the overall platform, check out our site and the presentation below, filmed at the Disney Accelerator demo day last year.