"Use loose quarterly OKRs alongside a shared north star metric"
Evidence from the Archive
Microsoft
Companies failing at AI: doing 'AI for AI's sake' with too many projects and no measurement blueprint
Microsoft AI Platform: north star metric plus loose quarterly OKRs across infrastructure, models, and agent tool chains
Oversees Microsoft's entire AI platform (infrastructure, foundation models, agent tool chains) while also having been COO of Instacart and VP at Meta -- giving her uniquely broad perspective on how goal-setting must adapt to both fast-moving AI development and traditional product operations. Their core argument: Use loose quarterly OKRs alongside a shared north star metric.
The evidence is specific: Microsoft AI Platform: north star metric plus loose quarterly OKRs across infrastructure, models, and agent tool chains. Furthermore, aI seasons: prototyping -> reasoning models -> agentic AI, each requiring different planning approaches. Instacart (previous role): consumer marketplace planning with more traditional OKR cadence for comparison.
In Asha Sharma's own words: "What is the north star metric is something that we do. The second thing that we do is that we have kind of loose quarterly OKR." (On combining north star metric with loose OKRs at Microsoft AI.)