AMA: Atlassian Principal Product Manager, Global Experiences, Liron Deutsch on Product Development Process

Always start with purpose. Why are you running the early access program? What assumptions are you testing, and what do you need to learn? Clarity here keeps your efforts focused and ensures the feedback you collect is actionable, not just interesting. Next, define success and exit criteria before launch. What does success look like, and what signals will tell you it is time to roll forward or roll back? This might include specific usage metrics, satisfaction thresholds, opt-out rates, or a set t ...Read More

This question made me smile. “Rogue engineers” is quite a term :) . First, I believe anyone, regardless of craft, should feel empowered to suggest ideas, challenge assumptions, and influence the roadmap. A healthy team culture thrives on curiosity and initiative. But if someone ships something outside agreed priorities without team alignment, it usually signals a deeper issue with how the team is operating. The best prevention is trust, transparency, and shared context. As PMs, we must constantl ...Read More

Changes in scope mid-cycle are common, and not always a bad thing. Sometimes they signal healthy learning, other times preventable misalignment. The key is knowing which is which and setting up governance that balances alignment with agility. First, prevent avoidable changes through my absolute mantra which is - continuous cross-craft collaboration. The most painful mid-cycle shifts I have seen came from teams not sharing enough context early and then being misaligned. As PMs, we need to constan ...Read More

I do not expect PMs to be analysts, but I do expect them to be data fluent. They should be able to self-serve, interpret results, and use data to make confident product decisions without always relying on a data scientist. At Atlassian, PMs are expected to be comfortable using analytics dashboards, building Amplitude charts, running experiments, and conducting pre- and post-analysis. When deeper statistical or modelling work is needed, they collaborate with data scientists, but the day-to-day pr ...Read More

Being at Atlassian, you can probably guess some of my answers, but tools are only as good as the habits that keep them current. Here is how we approach it.RoadmapFirstly, we keep our roadmaps up to date using Jira Product Discovery. And this is the best source of truth to also link and connect to all your other documents listed below. RequirementsWe capture requirements in Confluence, typically through a Project Poster or a PRD when more detail is needed. PRDs should always be paired with a prot ...Read More

Experience across multiple industries is a real strength for a PM. It shows adaptability, curiosity, and the ability to learn fast, all essential traits for product leaders. PMs who have worked across different domains often bring fresh perspective, pattern recognition, and a broader toolkit of approaches. They are usually quicker to identify transferable insights, connect dots between seemingly unrelated problems, and challenge assumptions that long-time insiders may not even see. When intervie ...Read More

At Atlassian, assessing which problems to tackle is a deeply cross-craft and strategic process. When we propose a new strategy or explore a meaty new problem space, it typically goes through a product review with leadership. These sessions resemble a pitch where the PM and team present a clear articulation of the problem, supported by customer research, market insights, and data analysis, and explain why this is the right problem to solve now. We are often asked questions like, “What would you d ...Read More

I like to think of backlog refinement, sprint planning, and strategic alignment as connected rituals that keep teams both grounded and adaptable. Backlog refinementThis should be a continuous process, not a one-off meeting. A healthy backlog is long enough that if priorities shift, the team can pivot to meaningful work straight away. It is usually a shared responsibility but primarily driven by engineering leads, who flag when the backlog is getting thin. Most teams I have worked with hold a for ...Read More

I advocate for shipping all experience changes as experiments whenever possible. This mindset keeps us learning-driven rather than output-driven. Every experiment starts with a clear hypothesis, defined success metrics (primary, secondary, and guardrail), and a section on what we want to learn and how we will learn it. When the experiment ends, we go back to those foundations. Did the results match our expectations? What did we learn? If it was not successful, why? These discussions are always c ...Read More