The Shared-Fate EM/PM Pair
Give the eng manager and PM the same performance rating, and make them understand before they argue.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 91%
Larson's diagnosis of EM/PM conflict is that it has exactly two causes: misaligned incentives, and — far more commonly — each side arguing before understanding what the other actually needs. He addresses the first structurally by rating the EM and PM as a pair on the same score (extended at Carta to a 'trifecta' including business leadership), and the second behaviourally by forcing understanding before any attempt at resolution.
Origin
Larson's practice as CTO at Carta, calibrated jointly with Carta's chief product officer. The extension beyond EM/PM to a broader 'trifecta' including business leadership comes from a blog post by Carta CEO Henry Ward.
Core principles
- 01There are no villains in the workplace — only people with complex incentives doing complex things.
- 02The genuinely hard cases are two good people whose joint execution isn't working, not one obviously bad performer.
- 03Shared rating means shared incentive: you're graded on solving the whole constraint set, not your functional slice.
- 04If you deeply understand what everyone actually wants, there is usually a compromise that gives everyone what they want without costing more time.
- 05What people say they want is often not what they actually need.
How to run it
- 1
Rate the EM and PM as one pair
Establish that the EM/PM pair generally receives the same performance rating, discussed in the same calibration meeting. At Carta, Larson and the chief product officer spend real time calibrating together.
Pro tip Consider extending to the trifecta — include the business leader, and for a design-led company like Airbnb, the design manager. Grade everyone on their ability to solve the entire constraint set, not just their functional constraints.
Watch out Keep an explicit exception for clear non-performance (an EM who can't show up, whose team doesn't respect them). Those cases are easy to diagnose and shouldn't drag a good partner down.
- 2
Tell people the rating is shared
The mechanism only works if it's known. Larson is explicit that people know this is what's happening because they've been told. Announced shared fate is what changes behaviour before the review, not after.
Watch out This makes incentives fair, not identical. The pair can still make bad trade-offs — they just can no longer profit individually from doing so.
- 3
Understand before you try to solve
Before pushing to ship the feature or change the approach, make sure you actually understand what the other person cares about. Dig past the stated position to the true need, which is often not what they're saying.
Pro tip The stated conflict is usually a proxy. 'We can't say yes to that timeline' and 'we're saying yes anyway' both hide needs neither party has voiced.
Watch out Skipping this step is the single most common cause of EM/PM friction Larson sees.
- 4
Surface the invisible constraint
Much friction comes from constraints the other side cannot see. The classic: an EM has been told by their team they need something new to work on, so they keep prioritising a rewrite the PM finds idiotic. The PM has zero visibility into that pressure, so the EM just looks like an unreliable partner.
Pro tip Ask directly: 'you keep prioritising this rewrite — what's going on?' You may still disagree, but now you can navigate the constraint honestly instead of narrating a villain.
Watch out The EM-must-supply-interesting-work assumption is itself worth questioning — it's the engine behind a lot of invisible constraints (see growth teams: fifty boring experiments versus engineers wanting to build something new).
- 5
Look for the no-cost compromise
With true needs on the table, look for the solution that gives everyone what they actually want. Larson's experience is that the 'hard trade-off' framing is usually premature — most conflicts dissolve into a compromise that doesn't take more time once you understand the real needs.
In the wild
Larson regularly hears EMs claim the PM only agreed to a sales commitment because the salesperson influences their promotion. The reality is simpler and less villainous: the business needs to sell things to remain functioning, and you can't just say no and expect the business to succeed.
→ Understanding the incentive replaces a victim/villain narrative with a real constraint the pair can negotiate. Larson's rule: the reality is never as simple as the story that finds a villain.
Larson and Carta's chief product officer calibrate EM/PM pairs together, giving the same rating on average, with exceptions only for clear individual non-performance. CEO Henry Ward has written about extending this to trifectas including business leadership.
→ EM and PM incentives are shared and known to be shared, which removes the structural reason to optimise against each other, even though the trade-offs remain hard.
Common mistakes
Casting your counterpart as the villain
People build simplistic narratives to find villains they work with. 'You won't believe what my EM partner did today' is the failure mode. There are no villains — just people with complex incentives doing complex things, and the villain story guarantees you never learn the real constraint.
Arguing the position instead of the need
Both sides push to ship, or push to change the approach, before either has understood what the other actually cares about. Since stated wants are usually not true needs, the argument is being fought over the wrong thing.
Assuming every conflict is a hard trade-off
The prevailing belief is that the field is full of unavoidable hard trade-offs. Larson's experience is that most dissolve once true needs are understood — a compromise usually exists that gives everyone what they want at no extra time cost.
Is it for you?
Best for
CTOs, heads of product and org leaders where EM/PM pairs chronically miss dates, blame each other, or optimise for their own function's review criteria.
Not ideal for
Cases of genuine individual non-performance, where shared rating would unfairly punish the functioning partner; also poor fit where a function has no stable counterpart pairing.
From the transcript
“empm pairs are Pairs and they generally have the same performance rating”
“before you try to solve the conflict it's like pushing to ship this feature pushing to change the approach just make sure you actually understand…”
“my experience is if you really deeply understand what everyone wants”
“there there are no villains in the workplace they're just people with like complex incentives”
From the episode
The engineering mindset
Will Larson (Carta, Stripe, Uber, Calm, Digg)