Outcome Ownership Over Role Boundaries
Goal people on the outcome, not the job description, and let them cross any function to reach it
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 90%
DoorDash's 'extreme ownership' is a concrete operating norm, not a slogan: a data scientist whose experiment failed picks up the phone and calls customers rather than filing a request with the qualitative research team. The mechanisms that sustain it are explicit — leaders model it, the goal is stated as the outcome rather than the role, everyone uses the product on a schedule (WeDash, four times a year), and the team is built as a net importer of talent from other functions so nobody's identity is a job title.
Origin
Jessica Lachs, applying a culture instilled by DoorDash founder/CEO Tony Xu and co-founders Stanley Tang and Andy Fang from the earliest days — including the sales lead handing out promo codes on Boston streets and the whole 20-person company answering support phones during an outage.
Core principles
- 01Own the outcome, not the box your role sits in
- 02Qualitative research beats quantitative when you need to know why — so go get it yourself
- 03Leaders must exhibit the ownership they expect and hire for it
- 04Structured product immersion builds empathy and surfaces bugs
- 05A team that imports talent from other functions cannot be precious about role boundaries
- 06Crossing into product, engineering, or ops work is not merely allowed — it is encouraged
How to run it
- 1
State the goal as the outcome, never the role
Tell people their goal is to figure out what's happening — full stop. If reaching it means picking up the phone and calling customers, that is what they do. The job title constrains nothing.
Pro tip Make this explicit on day one for every hire, whether they joined ten years ago or last month.
- 2
Model it from the top and hire for it
Ownership is both cultural and selective: leaders visibly do the unglamorous work, and the hiring process screens for people who want to win badly enough to do whatever is needed.
Pro tip Look for competitiveness and willingness to do work that isn't your job.
Watch out You cannot instil this by memo if leaders themselves defend their boundaries.
- 3
Reach for qualitative data when the numbers can't tell you why
When a shipped change fails and segmentation analysis can't explain it, stop and go talk to people. Have the data scientists themselves make the calls, because they know what question the next test needs answered.
Pro tip Whoever is blocked should gather the context — routing it to another team adds latency and loses the question.
- 4
Institutionalise product immersion
Put everyone into the product on a schedule. DoorDash's WeDash program has all employees go dashing or do customer support four times a year, building empathy with every side of the marketplace and surfacing bugs nobody would find from a dashboard.
Pro tip Pair up — Lachs goes pair-dashing with a colleague, which makes it a thing people look forward to.
- 5
Build a net-importer team of mixed backgrounds
Recruit into analytics from ops, engineering, marketing, and finance — not only career data scientists — and let people rotate out to product and ops too. Mix formal backgrounds (statistics PhDs, econometricians, economists) with consultants, operators, and financiers, and mix startup hustle with people who have seen scale.
Pro tip The payoff is mutual teaching: DCF models traded for statistics gotchas traded for kick-ass slides.
Watch out Technical bar still applies — imports must acquire the skills, on the job or in school.
In the wild
DoorDash shipped an affordability initiative that was expected to work and didn't. The team segmented the data and still couldn't explain why. Instead of handing it to the qualitative research team, the data scientists themselves sat and made phone calls to consumers to learn the motivations behind the behaviour.
→ The qualitative findings unblocked the team and shaped the design of the next test, which the data alone could not have specified.
In the 2014 Boston launch, the four-person team went out at 5am in winter to hand promo cards attached to KIND bars outside the T. The sales lead — compensated purely on signing merchants — was out there with them, because he wanted the business to win.
→ Extreme ownership became a durable cultural norm that Lachs still expects of her analytics team a decade later.
During a major site outage, the entire 20-person company jumped on customer support phones to issue refunds and rescue orders. Lachs, brand new and unable to use the tools, went out dashing to fetch pizza for the team so no Dashers would be diverted from customer orders.
→ One of the largest refund events as a percentage of the company's bank account — paid because the service had failed and it was the right thing to do.
Common mistakes
Defending the role boundary
'That's what the qualitative research team is supposed to do' is precisely the reflex that leaves a team blocked. If the work unblocks the outcome, it is your team's work.
Declaring ownership as a value without mechanisms
Extreme ownership at DoorDash is backed by leader modelling, hiring selection, a scheduled product-immersion program, and cross-functional talent flow. A values slide alone changes nothing.
Hiring only career specialists
A team of identical carbon-copy data scientists has no cross-teaching, less pragmatism about business impact, and stronger instincts to guard role boundaries.
Is it for you?
Best for
Founders and functional leaders trying to make 'extreme ownership' a real operating norm rather than a values-deck line, especially in specialist orgs like data science
Not ideal for
Highly regulated or safety-critical environments where role boundaries and separation of duties are controls, not bureaucracy
From the transcript
“yes you are a data scientist but you your goal is to figure out what's happening”
“the team data scientists included just sat and made phone calls”
“four times a year all the employees go out and go dashing or do customer support”
“we are a net importer of talent as opposed to a net exporter of talent”
“he was with us handing out promo codes because he was part of the team because he wanted to win”
From the episode
Building a world-class data org
Jessica Lachs (VP of Analytics and Data Science at DoorDash)