Layers of Product Management

Whoever you are competing with is probably a spreadsheet, an intern, or doing nothing

Beneath a request like preparing a tax return sit distinct benefits, weighed by importance and satisfaction

Specific enough to rule things out, or it is a mood rather than a vision

Where to compete, what to give up, and how to say so — decisions rather than documents

Solutions are more fun to discuss, which is why discovery usually stops one conversation too early

Waves of five to eight interviews, with the artefact revised before the next wave goes out

A problem can be real, frequent and painful and still be somebody else's to solve

Not a small version of the product but a measuring instrument with a falsifying result

Confidence should taper with distance, which is why Q4 cannot be drawn like next sprint

Scoring frameworks make a trade-off legible and quietly drop the opportunity cost

Capacity is not headcount: five people with two on call and one onboarding is not five

Five shipped features are a plan wearing a result's clothes, not a key result

Treat deletion as maintenance, because an item untouched for six months is telling you something

If the meetings happen but the loop still takes a quarter, you have the costume only

A spec both sides edit beats a spec one side presents, because handoffs are where products fail

Ship behind a flag when the code is ready, announce when the market moment is right

Attention is the scarcest input a launch spends, so decide in advance what success looks like

Which shelf you are on versus what the label says, and why relabelling cannot fix the shelf

Charge for something that grows with the value delivered and the number mostly takes care of itself

Fix the leak before opening the tap; borrowed tactics assume somebody else's retention curve

Name the decision each metric would change, and stop collecting the ones that fail the test

Weight feedback by the behaviour behind it: a daily user's complaint outweighs a non-user's suggestion

Set the threshold before the run, or a flat result turns into a directional win

Count the decisions your research produced rather than the studies you ran