How the Toronto Star Built an Engaged News Community

Two weeks ago, I attended the annual ONA Conference (it was my 9th time!). At conferences, I usually try to stay clear of sponsored sessions because (with some exceptions like Canva), they’re often just thinly veiled product pitches. With conferences becoming stingier due to shrinking budgets, the main attraction of sponsored sessions often seems to be food, happy hours and swag. The session sponsored by the Toronto-based audience engagement platform Viafoura wasn’t like that. No food, no drinks, no swag, just three genuinely interesting case studies. For this newsletter post I’m focusing on the Toronto Star case. 

Let’s dive in: 

Five years ago, the Toronto Star's audience engagement metrics were dismal. Today, the Star is outperforming competitors who have been building communities for much longer. This dramatic transformation was powered by artificial intelligence that automates content moderation, identifies story opportunities from reader conversations, and personalizes engagement at scale. The result offers lessons for any news organization looking to harness AI for authentic community building in an increasingly fragmented media landscape.

At it’s current scale, the Toronto Star reaches over 6 million readers weekly across print, desktop, mobile and tablet platforms with six daily news websites and 25 community websites.

The transformation centered on developing what Viafoura calls a "strategic community approach". What stands out to me: Contrary to many other news sites, articles are paywalled, but commenting is free. This might seem counter-intuitive and Star editors initially did have some trepidations about inviting off-topic comments from people who hadn’t read the article, but the AI-powered commenting system prevents that from happening. According to Viafoura, the commenting community now drives about a quarter of all registrations.

An innovative aspect of Toronto Star's community strategy is the implementation of Viafoura's "Story Agent"—an AI system that transforms community conversations into editorial opportunities using large language models trained on years of user-generated content.

How the AI Works: Every morning, editors receive an email with 6 different story pitches derived from the previous day's comment discussions. The AI analyzes conversations, comments and likes. It identifies interesting cluster groups and delivers pitches with specific angles via email directly to the newsroom via email. The AI specifically hunts for authentic human experiences. "One of the things that we're really trying to kind of tune this LLM for is to find personal stories and anecdotes," Viafoura's CEO Mark Zohar noted during the presentation.

The system identified 108 comments across 7 articles about dogs off-leash, synthesizing them into two distinct story angles that seemed most promising for ongoing engagement:

Each pitch includes highlighted reader comments that demonstrate personal stories and community sentiment—exactly the kind of authentic voices that drive subscriber engagement.

The practical application of this approach was demonstrated when Toronto Star's cultural reporter published a piece reviewing Toronto's best ice cream shops. The article generated 77 comments with readers sharing their favorite spots and discussing affordability.

Christine Loureiro, the Star’s Senior Editor for Digital Strategy and Initiatives, noticed the conversation brewing and posted: "Hi everyone, I'm an editor with The Star. We want to know about your spot for a summertime scoop. Tell us, and we'll share them with The Star readers in a future story."

Four days later, the Star published a follow-up story entirely based on reader recommendations, complete with clickable links back to original commenters' profiles. Featured readers received personal thank-you emails from the editor—the kind of relationship-building that drives subscription loyalty.

Beyond generating story pitches, Toronto Star's AI system serves as an on-demand research tool for reporters already working on assignments. When a transportation reporter queries the system about transit issues affecting Toronto residents, the AI scans community comments and suggests specific story angles, based on the comments on previous stories which express readers’ first-hand experiences and frustrations with Toronto’s transit system. 

These are some of the Toronto transit story ideas suggested by the bot: 

This approach enables new and more efficient ways for newsrooms to approach source development and story research. "Think about this as a workbench for a newsroom," Zohar explained. "You may have already identified a story you want to work on whether it came from story agent or not, but now you can go and drill down and say, okay, I'm working on this story. Find me some people who've experienced this problem." The AI effectively transforms months of community conversations into a searchable database of potential sources, story angles, and reader concerns—turning audience engagement into a competitive reporting advantage.

1. Use AI to mine comments for story ideas: Toronto Star's AI analyzes reader discussions and delivers six daily story pitches with supporting quotes and potential sources, turning community conversations into editorial intelligence.

2. Personal editor engagement converts readers to subscribers: When editors directly engage with commenters and send personal thank-you emails, they build loyalty relationships that drive subscriptions.

3. AI can serve as a community research tool: Reporters can query the AI system about specific topics to find story angles, sources, and reader concerns from past community discussions, making research more efficient.

4. Focus on personal stories and authentic voices: The AI specifically hunts for personal anecdotes and human experiences in comments, which creates more compelling content for many topics.

5. Strategic community approach beats basic commenting: Success came from developing comprehensive audience engagement rather than just implementing comment sections.