Shreyas Doshi on Product Management Frameworks
Shreyas Doshi in conversation with Lenny Rachitsky. Doshi shares five original frameworks from his career across Yahoo, Google, Twitter, and Stripe.
Key ideas
- Pre-mortem over post-mortem: before launching, prompt the team to imagine the project has already failed and work backwards. The prompt creates psychological safety for surfacing concerns that optimism culture suppresses — and builds a shared vocabulary (tigers, paper tigers, elephants) that persists into future meetings.
- LNO framework: all PM tasks fall into three types — Leverage (10–100× return), Neutral (roughly 1:1 return), and Overhead (return less than investment). Perfectionists by default treat all three the same; the fix is to direct deep effort only at L tasks and minimise time on N and O tasks.
- Three levels of product work — impact, execution, optics: conflicts between PMs and leadership often stem from each party defaulting to a different level, not a genuine disagreement. Naming the level makes the mismatch visible and fixable.
- Most execution problems are strategy or culture problems: persistent, recurring misalignments that bandaids cannot fix are usually caused by an absent or unclear strategy, a misaligned culture, or an unresolved interpersonal conflict — not by inadequate execution process.
- Minimise opportunity cost, not maximise ROI: in high-leverage roles, dozens of positive-ROI tasks exist at any time. Optimising for ROI pushes towards quick wins; optimising for opportunity cost forces asking “is this the best use of my time?” — a shift that surfaces the few transformative bets.
Pre-mortem
A pre-mortem is run at the start of a project, before anything has gone wrong. The facilitator opens with a single prompt: “Imagine this project has miserably failed six months from now — what went wrong?”
Three-category taxonomy for contributions:
| Category | Definition |
|---|---|
| Tiger | A genuine threat that could kill the project |
| Paper tiger | A seeming threat that others may worry about but the contributor does not |
| Elephant | Something not necessarily fatal but that nobody is openly discussing |
Process:
- Invite every function involved in the launch.
- Share the failure prompt; give participants quiet time to enter items into a shared document without seeing others’ entries.
- Go around the room; share all items.
- Ask each participant to identify the most concerning tiger that someone else raised — this functions as implicit voting.
- As leader, prioritise the identified risks and publish a pre-mortem action plan.
The vocabulary (tiger/paper tiger/elephant) persists into future team meetings without the leader present. Doshi considers this the most durable benefit.
LNO framework
Three categories of tasks in any high-leverage role:
| Type | Description | Effort-to-impact ratio |
|---|---|---|
| L (Leverage) | Tasks that return 10× or 100× the effort invested | High return |
| N (Neutral) | Tasks that return roughly what you put in | Roughly 1:1 |
| O (Overhead) | Tasks that return less than the effort invested | Low return |
The same type of activity can be any of the three depending on context. The L tasks a PM is not doing are usually causing the most anxiety — strategy work, difficult alignments, hard conversations.
Two tactics for tackling L tasks:
- Placebo productivity: spend two days before the appointed L task on intentional N and O work. The contrast creates the right mental state.
- Change of location: do the L task somewhere other than the usual desk.
Three levels of product work
| Level | Focus |
|---|---|
| Impact | What outcome does this create for users, customers, or the business? |
| Execution | What does it take to deliver this? How do we hit the next milestone? |
| Optics | How do we create internal awareness of what we are doing and achieving? |
Most conflicts between a PM and a CEO or founder are level-mismatch conflicts: the PM operates at execution level, the CEO at impact level. Neither is wrong — they are speaking different languages.
Optics is not inherently bad. The failure mode is when an organisation signals — through who it promotes — that optics alone is sufficient.
Most execution problems are strategy or culture problems
Persistent execution problems that resist repeated bandaids fall into three actual categories:
- Strategy problems: teams are misaligned because nobody has articulated a clear strategy.
- Culture problems: the organisation rewards individuals for hitting their own OKRs even when doing so prevents cross-team collaboration.
- Interpersonal problems: two team leads who cannot work together generate friction regardless of process.
Clear test: if the same problem recurs after a bandaid was applied, it is almost certainly not an execution problem.
Opportunity cost over ROI
When time is in the denominator, the natural optimisation strategy is to decrease it — pursue quick wins. This generates positive ROI in aggregate, but forecloses transformative bets.
Opportunity cost reframe: ask not “will this create value?” but “is this the best available use of my time?”
Practical implementation: use an allocation model as a planning heuristic (e.g. Google’s 70-20-10) — set rough percentages for incrementals, big new initiatives, and infrastructure, then let teams plan within those buckets.
Background
Shreyas Doshi was born in Mumbai, studied computer engineering in India. Worked as an engineer at several startups before transitioning into product management. Career spanned Yahoo (identity team), Google (six years, ads team), Twitter, and Stripe (led Stripe Connect). Began sharing product frameworks publicly on Twitter and built a large following.