The art of product management | Shreyas Doshi (Stripe, Twitter, Google, Yahoo)
Executive Summary
Shreyas Doshi joins Lenny Rachitsky to discuss pre-mortems, LNO task selection and the levels of product work. Selected coverage omits his background and early-Google overwhelm/stress anecdote. Company examples and claimed effects are his account, unverified here. Untimed topics are editorial, not video chapters.
Key Takeaways
Imagine failure before the launch
Doshi's pre-mortem starts by imagining that a project has failed and working backward to possible causes. It makes concerns easier to raise before the outcome is known. He still allows post-mortems for learning; the ritual is not a guarantee that nothing will go wrong.
Give risks a shared vocabulary
He calls a real harmful threat a tiger, a feared issue that you do not consider dangerous a paper tiger, and an unspoken concern an elephant. These labels can make a difficult conversation more concrete. They are categories for discussion, not proof that a concern has been validated.
Include the functions that will carry the risk
For a large launch he may separate engineering and go-to-market pre-mortems, with the involved people represented. Smaller launches can use one combined discussion. His quiet writing and sharing sequence helps collect concerns before a dominant voice frames everything.
Turn the discussion into an action plan
After people share, he asks them to identify a serious concern raised by someone else. The leader then prioritizes the output and owns follow-through. The purpose is not to solve every imagined problem but to address important risks that might otherwise be missed.
Match effort to leverage
LNO distinguishes leverage, neutral and overhead tasks by the impact returned for effort. Doshi argues against applying the same perfectionism to all three. The same activity can change category with context, so a bug report is not automatically low or high leverage.
Check what procrastination is protecting
He says difficult strategy or alignment work can be postponed because it needs focused attention and raises fear of doing it badly. That is his interpretation and personal experience, not a diagnosis of every busy PM. Identify one important task and protect time for it rather than treating all activity as equal progress.
Separate impact, execution and optics
Impact concerns the result; execution concerns getting the work done; optics concerns awareness of the work and result. Doshi sees useful internal communication as a source of feedback and energy, but warns against making appearances the goal. Teams can use the vocabulary to align on the level their context needs.
The Conversation Breakdown
Imagine failure before the launch
⏱ Untimed topicDoshi's pre-mortem starts by imagining that a project has failed and working backward to possible causes. It makes concerns easier to raise before the outcome is known. He still allows post-mortems for learning; the ritual is not a guarantee that nothing will go wrong.
Give risks a shared vocabulary
⏱ Untimed topicHe calls a real harmful threat a tiger, a feared issue that you do not consider dangerous a paper tiger, and an unspoken concern an elephant. These labels can make a difficult conversation more concrete. They are categories for discussion, not proof that a concern has been validated.
Include the functions that will carry the risk
⏱ Untimed topicFor a large launch he may separate engineering and go-to-market pre-mortems, with the involved people represented. Smaller launches can use one combined discussion. His quiet writing and sharing sequence helps collect concerns before a dominant voice frames everything.
Turn the discussion into an action plan
⏱ Untimed topicAfter people share, he asks them to identify a serious concern raised by someone else. The leader then prioritizes the output and owns follow-through. The purpose is not to solve every imagined problem but to address important risks that might otherwise be missed.
Match effort to leverage
⏱ Untimed topicLNO distinguishes leverage, neutral and overhead tasks by the impact returned for effort. Doshi argues against applying the same perfectionism to all three. The same activity can change category with context, so a bug report is not automatically low or high leverage.
Check what procrastination is protecting
⏱ Untimed topicHe says difficult strategy or alignment work can be postponed because it needs focused attention and raises fear of doing it badly. That is his interpretation and personal experience, not a diagnosis of every busy PM. Identify one important task and protect time for it rather than treating all activity as equal progress.
Separate impact, execution and optics
⏱ Untimed topicImpact concerns the result; execution concerns getting the work done; optics concerns awareness of the work and result. Doshi sees useful internal communication as a source of feedback and energy, but warns against making appearances the goal. Teams can use the vocabulary to align on the level their context needs.
Core Frameworks & Ideas
Pre-mortem
Imagine a failed project, surface risks across functions and prioritize an accountable action plan.
LNO
Distinguish leverage, neutral and overhead work, then allocate effort for the situation.
Impact, execution, optics
Clarify whether the conversation concerns outcomes, delivery or awareness before debating details.
Actionable Next Steps
- ✓Run a pre-mortem early with the functions involved in the launch.
- ✓Record real threats, apparent threats and unspoken concerns separately.
- ✓Assign an owner and next step for the most important risk.
- ✓Choose one high-leverage task and check whether status work is crowding it out.
Notable Quotes
“why do we need to wait until after things go wrong?”
“The main thing about a pre-mortem is to include people from every function that is going to be involved”
“it creates energy, it creates awareness, it creates excitement, it creates opportunities for feedback.”
The One Thing to Remember
Anticipate the risks and spend your best effort on work that changes the result, not only its appearance.