Most of what you're optimizing for is wrong
Shreyas Doshi spent years at Yahoo, Google, Twitter, and Stripe before concluding that a lot of standard PM behavior — careful execution, thorough status updates, ROI-ranked backlogs — is actively counterproductive. Not because those things are bad in isolation, but because PMs apply them indiscriminately, to the wrong problems, at the wrong level of abstraction.
Your to-do list is lying to you
Doshi's LNO framework draws a hard line between three types of work. Leverage tasks return 10–100× what you invest — a well-framed strategy doc, a hard conversation you've been avoiding, a pre-mortem that surfaces a project-killing assumption. Neutral tasks roughly break even. Overhead tasks consume more than they return — status emails, low-stakes reviews, meetings that exist because no one cancelled them.
The trap for ambitious PMs: perfectionism. When you bring the same standard of care to every task, you're implicitly treating a scheduling email like a product strategy. The correct move is to concentrate perfectionist energy on L tasks, do N tasks adequately, and do O tasks as fast as possible without guilt.
And here's the uncomfortable corollary: the L task is usually the one you're avoiding. Fear is a signal. The cave you dread entering is where the treasure is.
Naming the level of the conversation
Doshi argues that most team tension isn't disagreement — it's two people operating at different levels of product work without realizing it. He names three levels:
- Impact: Are we making the right product trade-offs?
- Execution: Are we building the thing correctly?
- Optics: Who knows what, and have the right people been communicated with?
A VP blindsided in a product review isn't questioning your impact decision — they're registering an optics failure. Until you can name which level a conflict lives at, you'll chase the wrong fix.
The level that's right also depends on context. Early-stage startups should default to execution — they haven't earned optics overhead yet, and impact questions are largely settled by founder conviction. As you scale, optics work grows (more teams depend on you) and impact work grows (the cost of being wrong gets bigger).
Execution problems are usually something else
When a team is persistently slow, the instinct is to add a new sync, create an escalation path, or hire a TPM. Doshi argues this is almost always a bandaid. In high-performing environments where everyone has good intentions, chronic execution failures trace back to strategy problems, interpersonal problems, or cultural problems. The bandaid masks the root cause; three months later it reappears as a new problem and gets a new meeting.
The honest diagnostic question: if this team has been 'slow' for six months, is the real constraint a missing standup, or is it a strategy so unclear that no one knows what 'done' means?
Prioritization is not ROI optimization
Doshi watched Patrick Collison at Stripe and drew a sharp conclusion: once you're senior enough that everything on your list has positive ROI, ROI stops being a useful filter. Your time is the constraint. A positive-ROI project that consumes a month forecloses the higher-impact project you didn't get to.
The correct filter at senior levels: minimize opportunity cost. Don't ask 'is this worth doing?' Ask 'is this materially better than the next-best use of this slot?' Most PMs never make this shift — they keep optimizing a measure that stopped being binding.
The pre-mortem as psychological safety technology
Doshi's pre-mortem ritual reframes a planning conversation using a single prompt shift: instead of 'what could go wrong?', the team imagines the project has already failed six months from now and works backwards. This framing gives everyone license to be negative — something modern organizations quietly discourage.
The output categories matter too: tigers (real threats), paper tigers (things people worry about but shouldn't), and elephants in the room (things nobody is saying but everyone should be). The pre-mortem isn't pessimism; it's structured honesty before the cost of honesty goes up.