Sonar's 25-Cent Solution to Spam Detection

Discourse and Sonar logos with a screenshot of Sonar community
“As soon as we turned it on, it was night and day.”

- Colin Mueller, Community Manager at Sonar, describing the impact of implementing Discourse AI to address a sudden and disruptive spam attack.

About Sonar

Sonar, the creators of SonarQube, have been helping developers write better code for over 15 years. They specialize in static code analysis for more than 30 programming languages. Sonar has developed over 6,000 rules to ensure code is reliable, maintainable, secure, sustainable, and accessible. Since 2018, Sonar has built an active support community using Discourse, leveraging its strong open-source roots to gather feedback, improve products, and provide support. 

We had the opportunity to chat with Colin Mueller, a Community Manager at Sonar, about their recent spam challenges and how Discourse AI features provided a swift and effective solution.

The Challenge: An unexpected surge in spam

Spam wasn’t a significant issue for Sonar’s community—until it was. In September, the community was hit with a wave of spam that bypassed existing spam tools. Members began to notice. 

Traditional mitigation strategies, like subnet blocks, were considered but posed risks. As Colin explained, “We knew, based on some data explorer queries, that we had very legitimate users coming from those subnets.” The spam also appeared to be human-driven, rendering CAPTCHAs ineffective. Manual approval wasn’t an option either, as it risked stifling community momentum.

“When you have a user who wants to help someone else in the community, manual approval really gets in the way,” Colin noted. 

The Solution: Discourse AI Triage feature

Faced with these challenges, Colin asked his CIO for permission to use Discourse AI. Once approved, he enabled Discourse’s AI Triage feature to tackle the spam surge. The system was configured to screen every new user post automatically through the spam detector. Posts flagged as spam were moved into a hidden holding area accessible only by admins.

This approach ensured moderators weren’t burdened with reviewing unnecessary alerts, allowing them to focus on meaningful interactions within the community. As Colin described, “Every time there’s a new user who creates a post, we run it through the spam detector. It decides whether it’s spam or not, and then we automatically move it to a holding area that’s hidden from everyone except for the admins.”

The Impact: A night-and-day transformation

The results were immediate and transformative. “As soon as we turned it on, it was night and day,” Colin shared. The community’s users saw the spam attack vanish as if it had never happened. The AI’s precision allowed admins to manage spam efficiently and transparently, maintaining the community’s momentum and trust.

Even more impressive is the cost. As Colin put it, “We did the math... on our busiest month, it’s 25 cents a month [for API tokens] for nearly perfect spam detection.”

Beyond resolving the spam issue, the experience sparked new possibilities. “Spam detection is just the beginning,” Colin said. “Large language models are made for processing written content, which is what community forums are all about. We can do so much more with this.”

Why Discourse?

Colin’s team has long appreciated Discourse’s flexibility and responsiveness of our support team. “It feels like every time you want to do something, there’s a toggle for that,” he remarked. This adaptability extends to AI features, allowing communities to choose their providers and model, track costs, and make prompt adjustments with ease.

“You’re not locked into any single provider,” Colin emphasized. “It’s super flexible. Just change a few words in the prompt or add another bullet point, and you’re good to go.”

Building long-term success

At Sonar, managing a vibrant and supportive community is important. Colin's goal is to make sure that the community members connect with the right resources "so that they can get the most out of the product… [but] when you're being just loaded by a spam attack, you can only focus on that." 

By addressing the spam surge effectively and transparently, the team at Sonar demonstrated how proactive tools can ensure a positive user experience for members while freeing up resources for more impactful work by community managers. 

With Discourse AI, the team is confident that the tools they need are just a toggle away. Our platform's thoughtful design and proactive features have solidified its role in how Sonar manages their community. “[Discourse] is a platform we plan to stick with for years,” Colin affirmed.

Conclusion

If your community is grappling with spam or looking for ways to automate routine tasks, Discourse AI offers a powerful, adaptable solution. We now offer an AI-based spam detector that can be set up in one click, delivering spam protection at no extra cost.

Learn more about Discourse AI and join our community