Finding actionable insights in audience signals you’ve been sitting on

One big thing we’ve had in mind since the NPAI Co-Lab was created: Newsrooms have qualitative insights everywhere.

Sometimes the sources are evident: open-ended survey responses, complaints to customer service. Other times, it’s less obvious what newsrooms’ available data can tell us about audiences. But signals can be found in newsletter replies or reader comments, too. Even the voice messages listeners leave can be valuable sources of data, as we saw in a pilot we recently ran.

How three newsrooms learned to unlock their qualitative data 

Collecting, managing, and leveraging first-party audience data — that is, information a newsroom gathers through direct interaction with its audience — is hard. As news organizations increasingly began using data to guide business and editorial strategy, that data ended up living in separate platforms, often disconnected from each other. It can be found in an analytics platform, an email platform, and a customer relationship management platform, among other disparate databases.

That’s how, even with so much data available, fragmentation keeps a newsroom from fully understanding its audience. The Audience Data Commons, an open-source schema created within the NPAI Co-Lab, gives newsrooms a shared, standardized way to structure that first-party data so it can be connected across sources and made actionable.

If fragmented data sources are already challenging to work with, unstructured ones are even harder. Yet that’s often where you can find insights that allow product and audience teams to measure behavior and trends, and optimize content and products.

In April, we asked three newsrooms to join us in exploring what could be possible if they unlocked the signals hidden across their unstructured qualitative data. Lookout Local, The Salt Lake Tribune, and Science Friday each brought us one or more first-party data sources and a question they wanted the data to answer. To do so, they could use the ADC schema and get engineering support for the setup from our partners at Newsroom Robots.

Over two months, we met and chatted regularly to compare notes on what the teams were finding. Sifting data through the schema fields, they were all able to surface qualitative signals they weren’t able to reach before.

• Science Friday found its most engaged listeners in voice-message transcripts that had been sitting in an inbox, read but never connected to each other.

• The Salt Lake Tribune grouped and made sense of hundreds of open-ended responses.

• Lookout Local discovered early signs of reader churn in data it already had, spread across various inboxes.

The pilot’s question was simple: Can the schema structure qualitative data and make it actionable? The answer was yes. And unlocking that is a starting point. Knowing you have fans, knowing who they are, and knowing where to find them is fundamental to any reader revenue strategy, especially for nonprofits that rely on people who believe in the mission. Structuring its open-ended responses let The Salt Lake Tribune make decisions about its ad strategy. Lookout Local now has the groundwork for an early warning system for churn, if it chooses to keep building in that direction.

None of these lessons were frictionless, of course. The pilot also served as a usability test for our shared schema assumptions. Early on, it surfaced how differently newsrooms might read a foundational term like “first-party data.” Some teams first reached for structured traffic and analytics data before we aligned on the qualitative, unstructured sources the pilot was meant to assess.

Mapping that messy, real-world data into a shared structure also took more work than expected. It demonstrated that a tool only helps a newsroom that is ready to use it. Some of the value newsrooms found in the schema depended on having a clear question going in, having the right access and permissions in place, or being able to export certain types of data at all. The schema could organize what was available, but organizations also needed to update their data mindset and management to make the best use of it.

The pilot gave NPA and Newsroom Robots the opportunity to help the participating teams understand their own audience data: what they had, what it could enable, and how to start organizing it. This goes beyond using the schema itself. We’ve been able to demonstrate that the schema works and makes first-party data actionable — especially for structured, strategic use cases — but a newsroom also has to be able to understand its potential. 

That leaves us with a few open questions. We’re still working on how to make interaction with the shared schema friendly for the people who work in audience roles and might not have a technical background. We’re thinking about how to scale this beyond answering a single question, to offer infrastructure that can support a newsroom’s entire strategy. Most important: How might we expand adoption of not just the ADC schema itself, but also the right posture to approach first-party data strategically?

We’ll share more details about the pilot experience soon. Until then, if you also have questions about infrastructure for unlocking and making the most of first-party data, join us in the #audience-data-commons channel on Slack. And as you explore the NPA Summit schedule, check out the NPAI Co-Lab hackathon, Connecting the Dots on Your First-Party Data, to find out how you can try this with us in person in October.