Lenny's PodcastSachin Kansal: Why Uber's CPO ships 300 fix-its every half
After 800 dogfood trips behind the wheel for Uber, he writes 40-page bug reports: every product team now ships 300 fix-its as a six-month OKR.
CHAPTERS
- 0:00 – 11:25
How Sachin dogfoods Uber at CPO scale (driving, delivering, documenting)
Sachin describes what “dogfooding” means at Uber across the rider, Eats, driver, and courier apps—down to personally driving/delivering 1–2 times per month. He explains how the real value comes from rigorous documentation (screenshots, write-ups, tagging owners) and following through until fixes ship.
- •Uses Uber and Uber Eats heavily as a normal user, plus regular hands-on driving/delivery sessions
- •Hundreds of real trips as a driver/courier to experience the product in context
- •Captures learnings with screenshots and detailed docs per app surface area
- •Tags responsible teams and follows through via an internal prioritization/fix process
- 11:25 – 15:40
Why it matters: emotion, context, and the limits of analytics
Sachin lays out a spectrum of user feedback—from metrics to surveys to focus groups to 1:1 conversations—and argues none replace the visceral reality of using the product while driving. He emphasizes that emotions (joy, outrage) create urgency and better decisions that pure numbers can’t.
- •User feedback spectrum: telemetry → surveys → roundtables → 1:1s → doing the job yourself
- •Driving changes everything: UI that looks fine on a laptop can fail at 45+ mph
- •Inefficiencies directly impact driver earnings, creating a sharper lens on problems
- •Emotion is a powerful internal motivator when paired with data
- 15:40 – 18:43
Empathy for drivers: the human interaction is the real ‘center of the universe’
A formative early driving story (the suitcase expectation) becomes a lasting empathy lesson. Sachin highlights that the driver-passenger relationship often matters more than app details and is under-discussed in office environments.
- •Surprising insight: in-car human interaction often dominates the driver experience
- •Early trip lesson: implicit expectations (e.g., luggage help) shape the job emotionally
- •Passenger mood and behavior can uplift or demotivate drivers
- •Dogfooding builds durable empathy that influences product decisions
- 18:43 – 20:06
How not to annoy your driver (and what drivers prefer)
Sachin explains that driver preferences vary (short vs. long trips) and offers practical etiquette that improves the experience for both sides. Small rider behaviors—like asking before taking a call or not slamming doors—can meaningfully affect driver sentiment and ratings.
- •Driver preferences are heterogeneous; incentives and personal constraints shape choices
- •Ask permission before being on the phone during the ride
- •Don’t slam the car door—it's a strong final impression
- •Rider behavior influences driver mood and ratings
- 20:06 – 22:04
Balancing metrics with experience: it’s not ‘data vs. empathy’
Sachin argues product teams need both funnel metrics and qualitative experience to build great products. The key is building a culture where emotion and numbers coexist, instead of letting optimization become a purely mathematical exercise.
- •Funnel metrics reveal friction; qualitative insights reveal lived reality
- •Numbers rarely capture in-the-moment feelings or urgency
- •Strong cultures allow “and” thinking: metrics + empathy together
- •Leadership’s job is to institutionalize that coexistence
- 22:04 – 24:26
Operationalizing dogfooding: company-wide weeks, OKRs, and rewards
Uber formalizes dogfooding via centralized enablement (how to sign up, regulatory constraints) and recurring “driving and delivery” weeks with friendly competition. Most importantly, feedback becomes a measurable commitment via fix-it OKRs (e.g., hundreds of issues addressed per half).
- •Central coordination helps scale dogfooding in a large company
- •Quarterly dogfooding weeks with competitions and prizes
- •Turns feedback into execution via tracked issues and fix-it OKRs
- •Reinforces behavior through expectations and rewards
- 24:26 – 26:22
Prioritization tradeoffs: funding fix-its alongside growth, retention, and cost
Sachin explains how Uber allocates resources across multiple OKRs rather than letting fix-its crowd out growth or vice versa. He emphasizes execution rigor, stretch targets, and leadership alignment to ensure quality improvements don’t get deprioritized to zero.
- •Fix-its are one OKR among several (growth, retention, cost savings, etc.)
- •Resource allocation is explicit during half-planning cycles
- •Execution rigor keeps projects moving; OKRs aim for ~75–80% attainment
- •Leadership buy-in is necessary to protect investment in experience quality
- 26:22 – 29:49
Dogfooding beyond consumer apps: merchants, admins, and “get as close as possible”
For harder-to-dogfood surfaces (merchant tools, Uber for Business), Sachin advocates creative proximity: shadow users, simulate flows with test accounts, and set up internal analogs (e.g., turning a campus barista into an Uber Eats merchant). The goal is always to stretch to the nth degree of closeness to the real customer workflow.
- •Uber Eats is a three-sided marketplace; merchant tools require different dogfooding tactics
- •Create internal testbeds (e.g., campus cafe as a merchant) to observe workflows
- •Spend time in restaurants and with large partners to see behind-the-counter realities
- •For B2B: shadow the admin + run the full setup yourself via test accounts
- 29:49 – 36:37
“Ship, ship, ship”: reducing cycle time by speeding decisions, not burning hours
Sachin connects dogfooding to execution: insight only matters if it ships as code in the product. He frames the enemy as cycle time—especially the dead space between real work steps—then shares mechanisms like product reviews, one-way vs. two-way door decisions, and occasional hands-on leadership interventions to unblock teams.
- •Impact comes from shipping code—not docs, meetings, or designs alone
- •Primary bottleneck is decision latency between discovery → design → build → launch
- •Use product reviews to force decisions; apply one-way vs. two-way door framing
- •Leaders step in selectively (e.g., daily standups post-COVID; writing a stake-in-ground PRD)
- 36:37 – 40:28
Why live demos: storytelling, launch rigor, and team pride
Sachin explains his insistence on live demos during product announcements, rooted in his Palm-era consumer electronics experience. Live demos both create external narrative clarity (people don’t inherently care about features) and internal rigor/pride by forcing products to truly work end-to-end.
- •External: live demos help tell a story that makes users care
- •Internal: demos enforce readiness and end-to-end quality before launch
- •Creating pride: teams see their work “real” on stage and in recorded history
- •Philosophy: users give you minutes of their life—earn their attention with clarity and value
- 40:28 – 44:04
Career advice: build product sense through repeated shipping and micro-decisions
Sachin advises early-career PMs to choose environments where they can ship repeatedly and quickly to develop judgment. He argues great PMs are made by thousands of micro-decisions (copy, layout, rollout choices), not a handful of big strategic ideas.
- •Pick roles with fast iteration cycles and multiple launches in 2–3 years
- •Don’t optimize for looking smart; optimize for shipping, learning, and iterating
- •Product excellence comes from many micro-decisions, not a few macro ones
- •Repeated cycles build “gut” and product sense faster than theory alone
- 44:04 – 46:55
How AI is changing PM work—and what stays constant (judgment and user understanding)
Sachin describes how AI accelerates many steps in the product cycle, from drafting PRDs to processing research and improving storytelling. He argues the enduring constant is deeply understanding users, and that judgment/product sense becomes even more important as AI commoditizes knowledge work.
- •AI is speeding up research, drafting, prototyping, and communication workflows
- •Constant across 24 years: truly understanding what end users want is still hard and critical
- •In an AI world, judgment and product sense become the differentiator
- •Leaders should further reduce cycle time to build faster learning loops
- 46:55 – 49:35
Engineers, PRDs, and boundaryless collaboration
Responding to a pointed question, Sachin shares what frustrates him most: teams being blocked waiting for a perfect PRD. He advocates co-creation—engineers, PMs, and designers whiteboarding together—and notes AI tools remove excuses for stalled documentation.
- •Dislikes “blocked until PRD” behavior; strong engineers can start with ambiguity
- •Prefers co-creation and shared ownership of the spec and solution
- •Operates boundarylessly: good product ideas can come from any function
- •AI can quickly turn brainstorms into draft PRDs, reducing friction
- 49:35 – 56:00
Uber’s autonomy strategy: partner ecosystem + a hybrid AV/human network
Sachin outlines Uber’s shift from building its own self-driving unit to partnering with many AV providers via APIs. The strategic bet is a hybrid marketplace where AV fleets gain utilization from Uber’s demand density, while human drivers fill gaps when AV supply can’t meet peak demand.
- •Divested internal AV unit during COVID; moved to a partner-led approach
- •~15 AV partners; integrates via a rich API layer into Uber’s marketplace
- •Hybrid network balances under-utilization (off-peak) and shortages (peak)
- •Autonomy also extends to delivery (e.g., sidewalk robots doing real volume)
- 56:00 – 1:01:58
Path to profitability: removing inefficiency while preserving growth bets
Sachin frames Uber’s profitability shift as rewarding but difficult, requiring innovation across batching, promotions, support, payments, and marketplace logic. He shares a portfolio model of concentric circles: protect and optimize the core at massive scale, then earn the right to expand into adjacent growth bets.
- •Profitability work involves thousands of efficiency projects (e.g., delivery batching)
- •Savings can be reinvested into users, drivers, and new bets; margins improve over time
- •COVID forced hard decisions (divestitures) and sharpened execution focus
- •Concentric-circle framework: flawless/efficient core → license to expand into adjacencies
- 1:01:58 – 1:10:26
When gut beats the data: safety sentiment, taxis, and Uber for Teens
Sachin shares several high-stakes decisions that weren’t clearly supported by data at the time. He emphasizes that if you understand the user problem deeply—especially through lived experience—taking principled risks can unlock meaningful new businesses and improvements.
- •Safety: building features that increase perceived safety, not just incident reduction
- •Taxis: data said declining/antiquated, but partnerships created real value (e.g., NYC yellow cabs)
- •Uber for Teens: large perceived risk, but a real parent pain point; strong adoption and demand
- •Principle: deep user understanding can justify contrarian bets
- 1:10:26 – 1:21:56
Failure corner + lightning round: resilience, paranoia about challengers, and practical tips
Sachin reflects on Palm’s slow response to iPhone/Android as a formative failure that taught urgency, resilience, and never taking the status quo for granted. The conversation closes with favorite books and shows, a life motto about focusing on inputs, and a final courier tip: turn on your porch light at night.
- •Palm’s decline taught urgency, hustle, and the difficulty of being #3/#4 in consumer markets
- •Maintains paranoia about innovative challengers and shifting strategies
- •Life motto: focus on controllable inputs over outputs
- •Courier tip: turn on the porch light to help safe, fast nighttime deliveries
