Metric-Owned Teams with Soft Feature Ownership
Give teams a metric to move, not a feature to own — anyone can touch any surface that helps their number.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 3
- Confidence
- 87%
Duolingo organizes retention-type teams around a metric (current-user retention rate, driving DAU) rather than around feature ownership. A team gravitates to whatever surface moves its metric best — retention 'owns' the streak only because the streak drives CERR better than anything else — but ownership is soft: any team may pitch and ship changes to any surface that helps their goal. This keeps effort on the highest-ROI work and avoids getting lost in subjective definitions of 'better'.
Origin
Described by Jackson Shuttleworth as Duolingo's team-structure philosophy; Lenny notes it mirrors how Airbnb's teams were organized around goals.
Core principles
- 01Own a metric, not a feature — it clarifies prioritization
- 02'Make the metric go up' is more actionable than 'make the feature better'
- 03Feature ownership is soft: whoever can move the metric may work the surface
- 04The metric owner acts as shepherd (bugs, roadmap) without a monopoly on changes
- 05Cross-team knowledge sharing beats territorial gatekeeping
How to run it
- 1
Assign each team a single metric to drive
Give the team a clear quarterly metric (e.g. current-user retention rate) rather than a feature charter.
Pro tip A laser metric focus makes it obvious what the highest-ROI work is; 'better' is too subjective to prioritize against.
- 2
Let the metric pull the team to its best surface
Allow the team to concentrate on whatever feature moves its metric most — for retention that happened to be the streak.
Pro tip Retention 'owns' the streak only because it drives CERR better than any other feature, not by decree.
- 3
Keep feature ownership soft
Let other teams pitch and ship changes to a surface you shepherd if it serves their metric; coordinate rather than gatekeep.
Pro tip The leaderboard team, or a learning team, can pitch a streak change; the streak team shares context and roadmap so the juice is worth the squeeze.
Watch out Because you likely hold multiple quarters of roadmap on the surface, incoming teams need your context to avoid conflicting or low-value changes.
In the wild
Jackson's retention team focuses on CERR/DAU, not on 'the streak' as a possession. They do most streak work only because the streak drives CERR better than any other feature, while other teams remain free to work on it.
→ Clear marching orders and prioritized backlogs; the streak gets the most iteration without being walled off from other teams.
Common mistakes
Organizing teams around feature ownership
'Make this feature better' is subjective and easy to get lost in; teams optimize the wrong things without a metric to anchor prioritization.
Treating a shepherded feature as exclusively yours
Hard ownership blocks other teams from high-value changes to a surface; soft ownership plus context-sharing keeps the best ideas flowing while avoiding conflicts.
Is it for you?
Best for
Growth-stage product orgs with clear north-star and driver metrics that want ROI-focused, non-territorial teams.
Not ideal for
Early teams without validated driver metrics, where feature-based ownership may still be needed for basic coverage.
From the transcript
“having a really strong degree of focus you know as as a team on you know what is the metric that I'm caring about and…”
“there's soft ownership we're not against teams doing things to the streak”
“we're trying to drive Kerr not we're trying to just make this feature better”
From the episode
Behind the product: Duolingo streaks
Jackson Shuttleworth (Group PM, Retention Team)