Sep 2, 2026
On working with PM (my favorite function!)
- pm is a uniquely difficult job because you're responsible for outcomes you don't control. This is hard on ppl. Good collab isn't possible without appreciation of this fact
- eng/pm strategic goals align but tactical goals diverge. This is by design. Thus the relationship is partly collaborative and partly adversarial
- strategically you're both trying to make the thing succeed. You may disagree on what that means or on how to do it, but agreement is possible at least in principle
- tactically you're operating on different planes. Pm wants to ship more features, or add capacity, or optimize latency. Engineering is overwhelmed with on-call, or is drowning in bugs, or struggling with deployments causing outages
- to make this relationship succeed each party needs to trust that the counterparty will work hard to find win/win outcomes
- sometimes tradeoffs really are zero sum and you have to negotiate a mutually acceptable outcomes. In this case good faith is even more important
Expectations:
- they should be trying to help you and you should be trying to help them, _especially_ in zero sum situations. If either one of you isn't doing that, you're not gonna have a good time. (This needs to be selected during interviews, it's very difficult to fix later)
- you should expect pm org to produce clear thinking derived from exhaustive inspectable research. Do not accept "PM'ing by vibes". You should be able to read user interview notes, inspect data, review facts and conclusions. “We spoke to Acme Corp and they said they want this” isn’t good enough
- on the flip side, do what you said you’re going to do. Ship super fast, high quality, scalable, reliable. Yes, this is hard, but that’s the job.
- relationship goes much better if you try really hard to understand their incentives, and openly communicate yours
- sometimes you have to push your team, sometimes you have to reject pm requests. In either case the counterparty won’t like it, and you’ve got to be ok with that
- if pm org is good you can learn a huge amount from them. It’s a tremendous learning opportunity. And if you’re both good they’ll hopefully be learning from you
- there are different kinds of pms— some tilt to data, some to user interviews, some to market research. I learned a lot from each
- at director level you need to agree on operating principles (like the above ^^) and work relentlessly to factually train your teams to work this way