Developer Advocacy: Building Trust, Measuring Impact, and Adapting to AI
An Interview with Mary Thengvall
Developer advocacy has evolved from a grassroots community movement into a strategic discipline that sits at the intersection of engineering, product, and marketing. But as organizations mature, many advocates still wrestle with fundamental questions: How do you measure impact? How do you scale community engagement without losing authenticity? And how will AI reshape how developers learn and adopt new tools?
InfoQ recently sat down with Mary Thengvall, a longtime practitioner and leader in developer relations and program management, to unpack these questions. Our conversation explored how advocacy has changed over the years, the balance between storytelling and metrics, and the new challenges AI introduces for both content and community.
The Accidental Advocate
Like many in DevRel, Mary didn’t start there intentionally.
“It wasn’t really a defined role when I began,” she recalls. “I was working at O’Reilly Media, asking questions like: Are we producing the right content? Are our conferences addressing what people actually need?”
Those questions evolved into what became the company’s first structured approach to community management—long before “DevRel” was a common title.
“I started by simply asking developers what they wanted to learn, what trends they were following, and what resources actually helped them. That curiosity—listening to the audience and responding with useful resources—became the foundation for everything that followed.”
Even now, in a program management role, that same mindset drives her work.
“Whether it’s building or managing strategic projects, the goal is the same: make sure people have what they need to succeed.”
Journalism, Empathy, and Product Thinking
Before tech, Mary studied journalism—something she credits with shaping her approach to advocacy.
“Journalism trains you to ask why people care about something, to find the human angle. That translates perfectly into developer relations. You’re still telling stories—but in a way that helps people solve problems.”
That empathy and curiosity also mirror product management principles: deeply understanding the user and their “jobs to be done.”
“If you can understand the human behind the code—their frustrations, motivations, and goals—you’ll always create better products and better content.”
Measuring Success Without Losing the Plot
DevRel’s toughest question is still: How do you measure success?
“It depends on what the company needs,” Mary says. “Early-stage teams might care about reach and awareness. Mature organizations focus more on retention, adoption, or reducing support load. There isn’t one metric—it’s about aligning to the company’s priorities at that moment.”
The hardest part, she adds, is resisting vanity metrics.
“A tutorial that addresses a common forum question might result in fewer questions about that feature, but can you prove causation? Usually not. We often deal with correlation, and that’s okay. What matters is that your work clearly supports business and user goals.”
That nuance, she says, is what often gets lost when DevRel is treated purely as a marketing function.
“When the focus shifts to traffic or impressions, you risk missing the point. Developer relations isn’t about noise—it’s about trust and sustained engagement.”
Marketing Goals vs. Developer Trust
Balancing authenticity with organizational goals remains a recurring challenge.
“Sales and marketing teams often focus on decision-makers, while DevRel speaks to practitioners—the people actually using the tools,” Mary explains. “I like to frame it as accounts versus individuals. Marketing targets accounts; DevRel supports individuals. Both are valuable, but the mindset is very different.”
Developers are quick to detect inauthenticity, so maintaining that distinction is essential.
“You build credibility by being useful. If the content genuinely helps developers solve a problem, they’ll remember that far more than a campaign slogan.”
Building Scalable, Authentic Content
When asked what types of content have proven most effective, Mary emphasizes diversity and reuse.
“People learn differently—some prefer videos, others prefer tutorials or live demos. I like creating what I call a content kit: start with a demo or sample app, document it, write a tutorial, record a walkthrough, maybe turn it into a talk or a podcast. You’re repurposing a single project into multiple formats.”
This approach expands reach without burning out the team.
“We measured what we could control. Our goal wasn’t always to publish—sometimes it was just to produce a content kit and hand it off to the team responsible for publishing. That kept expectations realistic while still showing progress.”
Lessons from Community Building
Community management is often the most rewarding—and the most difficult—part of developer advocacy.
“At Camunda, we ran both a Slack community and a Discourse forum. Slack was great early on—fast feedback, strong engagement—but as the community grew, it became unsustainable. The same questions came up repeatedly, and the answers weren’t searchable.”
The team made the difficult decision to shut down Slack and focus on Discourse.
“Engineers loved Slack because it felt personal. But long-term, a searchable, open forum was better for accessibility and sustainability. It wasn’t a popular decision, but it was the right one.”
That experience reinforced an important truth: sometimes the healthiest community decisions are the least popular ones.
“You can’t please everyone, but you can prioritize transparency and accessibility. That’s what lasts.”
Reporting Lines and Cross-Team Collaboration
For DevRel to be effective, it has to bridge multiple functions—engineering, product, marketing, and sales.
“I reported to the CTO, which gave me direct visibility into product discussions—roadmaps, feature trade-offs, customer pain points. That made it easier to anticipate questions from users and feed insights back to the team.”
At the same time, she stayed in sync with marketing and customer success to ensure alignment on messaging and outcomes.
“DevRel really sits at the hub of all those spokes. You have to build trust internally, not just externally.”
AI as an Assistant, Not a Replacement
With generative AI accelerating technical content creation, we discussed its role in DevRel.
“AI is amazing for overcoming ‘blank page syndrome,’” Mary says. “I’ll often use ChatGPT to draft an outline, summarize documentation, or generate a checklist. But you still need a human in the loop to verify accuracy and context.”
She once used AI to condense complex migration documentation into a concise checklist—then had engineers review it for correctness.
“It saved time, but it didn’t replace expertise.”
The Future of Discovery and API Adoption
Developers are already using AI tools to interact with documentation and troubleshoot APIs. Mary believes this trend will reshape how teams think about discoverability.
“It’s a bit like SEO. AI assistants pull from what’s already well-structured and visible online. So if your docs and examples are clean, discoverable, and current, your product will show up in those answers.”
Right now, though, she sees AI affecting how developers learn more than where they learn.
“People might skip the forum and just ask an LLM—but that LLM is still pulling from those forums. The consumption pattern changes, but the underlying knowledge sources don’t—at least not yet.”
What Makes a Great API Experience
When it comes to adoption, Mary believes the fundamentals haven’t changed.
“A great API is intuitive. It gets developers to value quickly and minimizes confusion. Every additional question—‘What does this parameter mean?’ ‘Why isn’t this call working?’—adds friction.”
The best developer experiences, she says, prioritize simplicity and fast feedback loops.
“Clear documentation, quick-start guides, sample apps—those are still the biggest levers. Anything that shortens time-to-value increases trust and retention.”
Avoiding an AI Echo Chamber
Looking ahead, Mary sees both promise and peril in how AI will shape developer marketing and content.
“The risk is that we stop creating new ideas. These models learn from the past, so if we’re not careful, we’ll just keep remixing what’s already out there.”
The real opportunity, she says, lies in pairing AI efficiency with human originality.
“The people who thrive will be the ones who use AI to accelerate workflows—but still bring their own insight, creativity, and lived experience. That’s what will set future content apart.”
Transitioning from DevRel to Program Management
Today, Mary is a program manager for strategic projects at Camunda, coordinating cross-functional initiatives across the company.
“It’s less about direct advocacy and more about ensuring teams have the resources and feedback loops they need,” she explains. “But the mindset is the same—help people succeed, remove friction, and communicate clearly.”
That through-line—from journalism to community management to program leadership—reflects what might be DevRel’s greatest lesson: the importance of curiosity and empathy in every technical discipline.
“At its core, developer relations is about listening and connecting. Whether you’re writing code, managing programs, or creating content, that human focus never really changes.”