Lenny's PodcastJessica Lachs: Why centralized data beats embedded analysts
Through pods aligned to product and ops, DoorDash centralizes data; analysts share goals with their teams, and case interviews hire for curiosity over skill.
CHAPTERS
- 0:00 – 4:59
Jessica’s DoorDash journey and what “analytics as impact” really means
Lenny sets the stage for why DoorDash’s data org is widely admired, and Jessica frames analytics as a decision-driving partner—not a ticket-taking service. She explains the expectation that analytics answers not only “why,” but “so what” and “what should we do now.”
- •DoorDash as a complex multi-sided marketplace that demands strong analytics
- •Analytics should drive business impact, not just reporting/dashboarding
- •The “seat at the table” mindset: insights + recommendations
- •“Answer the so-what” as a core team expectation
- 4:59 – 11:05
Centralized vs. embedded analytics: definitions and Jessica’s strong stance
Jessica explains what she means by a centralized model versus embedded analytics, focusing on reporting lines while keeping goals aligned with partner functions. She outlines why business leaders often prefer embedded teams, and how DoorDash preserves those benefits without losing centralization advantages.
- •Centralized = analytics reports through a central data org; embedded = reports into functions
- •Goals can still be shared with marketing/product/ops even in a centralized model
- •Why embedded feels attractive: camaraderie, control, roadmap certainty
- •DoorDash’s compromise: centralized reporting with functional “pods”
- 11:05 – 15:10
Why centralization wins: talent bar, growth paths, and consistent definitions
Jessica details the practical benefits DoorDash gets from a centralized analytics “center of excellence.” She highlights how it improves hiring consistency, creates stronger career mobility, and prevents metric and methodology fragmentation across the company.
- •Consistent, high talent bar via shared rubrics and evaluation standards
- •Better career growth and retention through internal mobility
- •Shared metric definitions prevent conflicting numbers across teams
- •Reusable methodologies/models reduce duplicated work and improve scalability
- 15:10 – 17:23
Protecting proactive work: balancing deep dives with constant inbounds
The conversation shifts to a universal analytics challenge: making space for exploratory, high-leverage work while still answering day-to-day questions. Jessica shares tactics DoorDash uses to intentionally carve out time and maintain accountability for proactive insights.
- •Exploratory work is the first thing to disappear without intention
- •Set explicit team goals for self-directed insights and deep dives
- •Use structured time blocks like hackathons to force exploration
- •Partner buy-in grows when deep dives demonstrably shape roadmaps
- 17:23 – 20:21
Hackathon case study: referral channel deep dive and fraud discovery
Jessica shares a concrete example of proactive analytics work that paid off. A referral analysis revealed the channel’s performance was “bimodal,” with valuable users mixed with fraud and low-quality acquisition—leading to policy and product changes.
- •Referral looked weak on average, but the average was misleading
- •Hands-on investigation (even attempting fraud) uncovered real exploits
- •Bimodal distribution: great referrers vs. discount-seeking/fraud behavior
- •Recommendations: stronger fraud checks, caps, and smarter channel optimization
- 20:21 – 24:17
Pushing back effectively: trade-offs, prioritization, and goodwill
Jessica explains how data teams can say no without being adversarial. The key is making trade-offs explicit, anchoring on shared goals, and creating a regular prioritization cadence with cross-functional partners.
- •Use shared goals to align priorities with partner teams
- •Make trade-offs visible: “If we do this, what drops?”
- •Avoid silent over-commitment; turn it into a prioritization conversation
- •Occasionally do quick requests to build goodwill—but don’t let it dominate
- 24:17 – 28:56
Hiring for curiosity and real-world problem solving (beyond technical skills)
Jessica explains what separates great data hires from merely capable ones: curiosity, self-motivation, comfort with ambiguity, and the ability to form a point of view. She shares how DoorDash interviews for these traits through realistic cases.
- •Technical competence is table stakes; curiosity is the differentiator
- •Interview tactics: embed “something off” in cases and see who notices
- •DoorDash uses DoorDash-history cases to test structured thinking
- •Signals: how candidates react to being wrong, incorporate new info, and decide
- 28:56 – 34:40
Non-traditional background: self-teaching data skills and leading with pragmatism
Jessica describes becoming a data scientist out of necessity, learning SQL/Python to solve immediate problems. She argues her finance and operator background helps keep highly technical teams grounded in business impact.
- •Self-taught analytics driven by startup needs and problem urgency
- •First-principles focus: solve what’s in front of you; don’t over-plan the org
- •Leverage complementary strengths: hire deep technical experts, lead with pragmatism
- •Business impact as the north star for sophisticated analytics work
- 34:40 – 38:20
Early DoorDash culture: extreme ownership and customer-first behavior
Jessica shares formative stories from launching Boston and from early HQ crisis moments. The theme is extreme ownership—everyone did whatever the business needed—and a deep commitment to customers, even when it was costly.
- •Boston launch: scrappy growth (5am promos), whole-team effort beyond job scopes
- •Ownership story: sales helping acquisition because winning mattered
- •Early HQ outage: entire company jumped into customer support
- •Customer-first refunds and doing “the right thing” even when expensive
- 38:20 – 44:17
Institutionalizing ownership: WeDash and encouraging cross-functional behavior
The conversation turns to how DoorDash scales early culture into a bigger company. Jessica describes WeDash and gives examples of data scientists doing qualitative research and stepping into product/ops work to unblock outcomes.
- •WeDash program: employees dash or do support multiple times per year
- •Build empathy across consumers, Dashers/couriers, and merchants
- •Data scientists making customer calls when quant alone can’t explain outcomes
- •Cross-functional exploration enables internal transfers and stronger operators
- 44:17 – 50:56
Defining effective metrics: proxies, simplicity, and a common business currency
Jessica lays out her philosophy for metrics: pick short-term proxies that drive long-term outcomes, keep metrics understandable, and translate initiatives into a shared “currency” for decision-making. This enables faster, clearer trade-offs across teams.
- •Avoid goaling purely on long-term outputs like retention; use proxy inputs
- •Prefer simple, intuitive metrics over opaque composite scores
- •Translate levers (price, delivery time, selection) into a common unit (e.g., orders/GOV)
- •Use shared currency to compare investments across marketing, logistics, supply, and product
- 50:56 – 55:28
When metrics get too complex: the ‘merchant health score’ simplification lesson
Jessica shares an example where a composite metric became unusable because nobody could interpret or move it. The fix was breaking it into a small set of clear input metrics and a straightforward outcome metric for activation.
- •Composite metrics can become meaningless if teams can’t interpret changes
- •.35 vs 35% problem: unclear magnitude and directionality of improvement
- •Replace one complex metric with a few high-signal, actionable metrics
- •Focus on the top drivers first (the 95%), then iterate toward the remaining 5%
- 55:28 – 1:00:11
Edge cases and fail states: why rare “disaster” events deserve dedicated goals
Jessica argues that averages hide the experiences that cause the most damage. DoorDash explicitly metrics and targets rare but costly failures (like “never delivered” orders) because they drive churn and expensive remediation.
- •Averages mask rare but severe customer experiences
- •‘Never delivered’ is rare but highly churn-inducing and expensive
- •Create explicit goals to eradicate fail states, not just improve averages
- •Also consider missing data: some failures (e.g., login issues) don’t show up in datasets
- 1:00:11 – 1:02:31
Managing a global data org: what changes (and what surprisingly doesn’t)
Jessica discusses running analytics across geographies and brands (including Wolt). While regulations, currencies, and languages add complexity, she’s struck by how similar both people and marketplace dynamics often are across countries.
- •Global expansion adds complexity: currencies, languages, EU vs non-EU regulations
- •Many consumer/courier/merchant behaviors are more similar than expected
- •US lessons often transfer—like having “the answer key”—but still require testing
- •Differences are valuable surprises that keep strategy and analytics honest
- 1:02:31 – 1:08:40
AI for productivity: scaling help with ‘Ask Data AI’ and self-serve analytics support
In AI Corner, Jessica focuses on using AI to make teams more productive and reduce analytics bottlenecks. She describes evolving long-running “Office Hours” into AI-assisted support so non-technical teammates can improve queries and answer questions independently.
- •Office Hours began as a manual weekly support mechanism for bandwidth constraints
- •AI opportunity: help edit/transform SQL and guide analysis without direct analyst time
- •Goal is broader empowerment of non-technical users, not just analyst efficiency
- •Internal tool: ‘Ask Data AI’ (clarity-first naming)
- 1:08:40 – 1:19:55
Lightning round and closing reflections: influences, habits, and DoorDash’s inflection moments
Jessica shares personal recommendations and the people who shaped her career, plus a sleep-centric problem-solving motto. She closes with moments when DoorDash’s success became undeniable, and where to follow her writing.
- •Books/entertainment: historical fiction, spy genre, The West Wing rewatches
- •Life motto: ‘committee of sleep’—sleep improves clarity and decisions
- •Career influences: standout women leaders, and parents’ example of reinvention
- •DoorDash inflection moments: becoming #1 in market share and seeing widespread product usage
