AMA: Atlassian Director of Product Management (Confluence), Paresh Vakhariya on Platform Product Management

How I actually approach alignment: Get upstream Platform PMs need to be in planning cycles early quarterly roadmap reviews, annual planning, OKR-setting. If you're only showing up when there's a conflict, you've already lost. Make the platform the path of least resistance If adopting your platform abstraction is genuinely faster than building from scratch, product PMs will choose it without being asked.  Explicit dependency and adoption mapping Every quarter I maintain a simple map: which platfo ...Read More

A Platform PM owns the product layer that other teams build on top of not the end-user experience, but the infrastructure, APIs, shared services, and developer tooling that make those experiences possible. Define and own the platform's contract. This means APIs, data models, authentication patterns, permissions, and the guarantees you make to internal and external consumers. Manage internal customers like external ones. Your stakeholders are engineering teams, other PMs, and sometimes partners. ...Read More

A feature PM succeeds when users adopt and love a specific capability. A platform PM succeeds when other teams ship better products faster because of what you built.  Adoption and usage by internal consumers Are teams choosing to build on your platform, or working around it?  Time-to-ship for downstream teams Before/after benchmarks on how long it takes a product team to launch a capability your platform now handles.  Reduction in duplicated systems How many one-off implementations of the same c ...Read More

Four signals I would use: Duplication When I see the same capability being rebuilt across multiple teams, that's the clearest signal of a platform gap. e.g. Identity, Notifications etc. Frequency of the ask, not volume of the noise I track how many distinct teams are asking for the same thing, not how loudly any one team is asking.  Tax on shipping What's slowing teams down right now? If authentication, permissions, data access, or observability keep appearing as blockers that's where platform i ...Read More

Here's how I would think: The abstraction boundary problem The hardest architectural decision in third-party integration is deciding how much of the third-party model to expose internally. Expose too much and your platform becomes tightly coupled to a vendor's data model. Abstract too much and you lose the capability that made the integration worth building. Reliability and SLA inheritance When a third-party goes down, your platform goes down and your internal consumers blame you, not the vendor ...Read More