The problem I was hired to solve
A meaningful share of Sonar's growth depends on customers migrating from on-prem SonarQube Server to SonarQube Cloud — and on enterprises being willing to connect AI capabilities safely into their toolchain. Both were stalling. The question was: why, and what would it actually take to unblock them?
What I did
- On-prem + cloud DevOps discovery: Researched customers who were running SonarQube Server on-prem while simultaneously using cloud-based DevOps tooling. The goal was to understand why their setup was split, what specifically was blocking migration to SonarQube Cloud, how remediation happened in their current environment, and what agentic remediation would need to look like to be viable for them.
- Integration persona development: Developed three integration personas — the Human Orchestrator, the Risk Manager, and the Autonomous Agent — each with distinct goals around setup, trust, visibility, compliance, and the degree of autonomy they were willing to grant to AI systems.
- AI persona evolution work: Led a broader persona study on how AI is reshaping user roles and needs across the organisation. Examined how enterprise customers were adopting AI through different operational models and what each model required in terms of controls, visibility, and coordination.
What I found
Migration wasn't stalling because SonarQube Cloud lacked features. The real blockers were organisational and commercial:
- Migration effort: Large customers had deeply embedded SonarQube Server configurations. Moving wasn't a technical toggle — it required internal approvals, security reviews, and months of coordination.
- Pricing history: Many long-standing customers had legacy pricing that made the cloud equivalent feel like a price increase, regardless of the feature set.
- Security and data residency: For regulated industries, where code analysis ran and where data lived wasn't a secondary concern — it was a deal-breaker if not addressed explicitly.
- Proving the value of the move: Customers couldn't easily quantify whether the disruption was worth it. They needed evidence, not feature lists.
On the agentic side, the persona work revealed a more nuanced picture: enterprise adoption of agentic workflows wasn't a single decision — it was two linked but distinct problems. Orchestrating value (getting the agent to do useful things across the toolchain) and governing risk (permissions, auditability, cost control, and oversight at scale) required different solutions for different personas. Sonar would need to support multiple modes of AI-enabled development, not assume a universal user model.
The decisions this enabled
- Reframed the migration problem internally: not a product-comparison problem, but an organisational adoption problem — which required a different GTM and product response.
- Gave the team sharper, more credible language for the barriers customers actually faced — language that could be used in customer conversations, sales plays, and product roadmap justifications.
- Created a persona model that made governance, trust, and operational fit explicit parts of the product strategy — not afterthoughts.
- Strengthened the through-line between migration research and the longer-term Agentic SDLC vision, making it easier for leadership to see how near-term and long-term work connected.