Classifying Engagement: Hearken Wins Funding from the GNI Innovation Challenge

Jennifer Brandel
We Are Hearken
Published in
4 min readDec 15, 2020


Hearken has worked with more than 250 organizations around the world in the past five years, and we’ve seen a tremendous variance in sophistication with what can be broadly called “audience engagement.”

We’ve also seen that engagement has become central to our newsroom partners’ strategy, but hasn’t yet been widely realized as a viable business model foundation.

But we know that it is.

In newsrooms that have connected databases and staff who are paying attention, we see incontrovertible evidence that more audience engagement with editorial leads to better business outcomes.

Smartly Classifying Audience Engagement For Newsroom Action

Imagine you’re a newsroom that’s in the swirl of a major news event — it may be a natural disaster (wildfires, earthquakes, hurricanes, tornadoes, extreme temperatures, a pandemic) or a man-made crisis (mass shootings, riots, dismantling democracy), and you know that the people you serve are anxious. They need answers, and fast, in order to determine how to make decisions about their lives, their families and communities. They’re looking to their trusted news sources to help them navigate the unknown.

In this situation, in which every newsroom has found itself in, and increasingly often, you’d probably look to engagement methods to hear what your audience needs. But what if you get way more input and demand for information than you can possibly handle? You don’t want to leave your anxious public hanging, and you also can’t afford to have staff working non-stop.

The above scenario is one we’re seeing newsrooms facing all the time. In fact, when we looked across all of our newsroom partners between the months before and after COVID-19 took hold, we saw an eye-popping 1000%+ growth in audience questions over a period of 3 months.

So we took action to support them in making sense of the inquiries and answering as quickly and effectively as we could. Leveraging Natural Language Processing capabilities on audience engagement flowing into Hearken’s newsroom partners, we tested our product’s ability to better organize audience information needs into actionable insights.

We tested this approach out with a model built by Quartz AI studio (shout out to John Keefe and Jeremy Merrill) for KPCC + LAist (shout out to Ashley Alvarado), a public media newsroom in Los Angeles, to thousands of questions they’ve received about COVID-19. Through teaching a machine to classify the questions, we dramatically cut down on the human hours it took the team to sort, identify duplicates and trends, and ultimately respond to people.

We know that engagement is highest in times of need, and KPCC is finding that taking this targeted, customer-service approach to their journalism is resulting in more newsletter subscribers, more loyal and local audience interaction, and more membership.

To train our algorithm, we have 5+ years of submissions from Hearken partners around the world. This enables us to automatically classify topics with a high degree of accuracy and surface information needs and trends that are all but impossible to notice through human sorting alone.

What’s In It For Hearken Newsroom Partners

The funding from Google News Initiative will enable Hearken to build features into our Engagement Management System that:

  • Enables journalists to move into the proactive end of big stories, rather than reactive
  • Saves newsroom time and money through efficiently responding to and providing new content for novel information needs (help more people with fewer newsroom resources)
  • Identifies trends and actionable insights that would otherwise go unnoticed
  • Incentivizes newsrooms to pay more and closer attention to engagement
  • Helps editors specifically and directly tie their coverage decisions to listening to the public
  • Helps newsrooms be as relevant and as worthy of financial support as possible

Ultimately, we want to make it as easy as possible for the information needs of the public to be surfaced, organized, acted upon and responded to by newsrooms so that citizens can do their jobs and make informed decisions in their communities.

In addition to the Engagement Management System, Hearken also has a Community Management System technology, which acts as a mutual aid network. We can then create integration between people’s information needs and the manifestation of the change communities are seeking to make, through weaving together insights from these two technologies.

Curious? Here’s How to Get In On This

For Hearken partners, we’d love to add you to our early beta testing group! Fill out this brief form to indicate your interest and we’ll be back in touch.

For not-yet Hearken partners, reach out to us here. We’ll look forward to learning more about what you’re interested in and seeing if we can support you!

We’re looking forward to sharing our learnings with you as we go on Medium and through our free monthly newsletter, The Hearkening. You can also check out our product roadmap for the Engagement Management System here and for the Community Management System here, to see what we’re working on.

And if you happen to be a Data Scientist / Machine Learning Engineer — we’re hiring that role to support this project! See the position details and apply here.

Special thanks to Hearken’s VP of Product, Kavya Sukumar, who dreamt this up alongside Julia Haslanger.

Check out the other 32 projects Hearken is being funded alongside, and all of the GNI-funded projects here.



Jennifer Brandel
We Are Hearken

Accidental journalist turned CEO of a tech-enabled company called Hearken. Founder of @WBEZCuriousCity Find me: @JenniferBrandel @wearehearken