Product lessons from Waymo | Shweta Shrivastava (Waymo, Amazon, Cisco)
Shweta Srivastava, Senior Director of Product Management at Waymo, discusses building trust in self-driving cars, Waymo's KPIs, and lessons from her career at companies like Amazon and Cisco, emphasizing customer focus and self-disruption.
Deep Dive Analysis
13 Topic Outline
Introduction to Shweta Srivastava and Waymo's self-driving service
Shweta's team responsibilities at Waymo
How Waymo builds trust and communicates with riders
Product management differences at Waymo vs. software companies
Waymo's commercial and system behavior metrics
Understanding L4 and L5 autonomous vehicles
Keeping investors engaged in long-term projects
Lessons for building successful products and teams
Importance of defining what not to build
Why large companies must disrupt themselves
Underrated product management skills: listening and empathy
Advice for career promotion
How to try Waymo's ride-hailing service
5 Key Concepts
L4 Autonomous Vehicles
L4 refers to fully autonomous driving without a human driver at the wheel, with no expectation of human intervention. Waymo focuses on building L4 systems, which operate reliably in specific, defined operational design domains like mapped cities.
L5 Autonomous Vehicles
L5 represents full autonomy in any driving environment, including unstructured roads, off-roading, and without reliance on maps or prior data. Shweta suggests L5 might become a niche solution, as L4 can largely fulfill the dream of autonomous driving.
Working Backwards (Amazon Process)
This product management approach, popularized at Amazon, requires product managers to write a press release for a finished product before development begins. This process forces a thorough consideration of the product's value proposition and customer problem it solves.
Innovator's Dilemma
This concept describes how large, successful companies can fail by focusing too much on incremental improvements for existing customers, making them vulnerable to disruption by new technologies or business models from smaller, agile competitors. Product leaders should proactively disrupt their own models.
MVP Bar for Safety at Waymo
At Waymo, the concept of a Minimum Viable Product (MVP) has a uniquely high bar due to the critical importance of safety. Unlike traditional software where rapid iteration is common, Waymo cannot cut corners on safety, making the initial MVP for autonomous driving extremely robust.
8 Questions Answered
Waymo builds trust by training deep-learned models on human driving data to ensure the car's behavior is natural, predictable, and safe, mimicking good human driving. They also incorporate social norms into the system and provide transparency through in-car monitors and human support.
PMs at Waymo must go technically deeper, be comfortable with high uncertainty and a long-term game, be driven by the mission of road safety, and navigate an extremely high MVP bar for safety, which means less room for cutting corners compared to typical software products.
Waymo tracks progress through commercial metrics (e.g., trips per week, active users, operational costs) and system behavior metrics. The latter includes safety (aiming to drive safer than humans), compliance with road rules, and the ability to make adequate progress without undue stops or stranding.
L4 is fully autonomous driving without a human driver at the wheel in specific, mapped environments, which is Waymo's current focus. L5 would be full autonomy in any unstructured environment without maps or prior knowledge, which Shweta believes might be more niche.
To maintain investor enthusiasm, companies must show meaningful progress in both technology and commercial deployments. The focus should be on building a viable business that creates value for customers, demonstrating tangible results rather than seeking short-term gains.
Listening and empathy are underrated skills for product managers. It's crucial to approach situations with a beginner's mindset, absorb information, learn, and proactively challenge one's own assumptions to truly understand customers and market needs.
Shweta advises focusing on creating significant impact for the business and doing what is right for the company, rather than optimizing actions solely for promotion. It's important to make ambitions known to your manager and seek challenging, high-visibility projects that stretch your skills.
Waymo's service is open to the public in the Phoenix metro area and San Francisco; users can download the app to use it. Phoenix currently has no waitlist. Waymo is also expanding its service in Los Angeles throughout the rest of the year.
25 Actionable Insights
1. Work Backwards from Customer
Always work backwards from the customer or user problem, focusing on what you are building, who it’s for, and what problem it solves, rather than building technology for its own sake.
2. Disrupt Yourself Proactively
Challenge your own models and product capabilities to disrupt yourself proactively before an upstart does, as this constant self-disruption is key for long-term success, especially in large companies.
3. Define What Not to Build
Clearly define what you are not building to maintain focus and prioritization, avoiding the trap of trying to be all things to all people, which dilutes product effectiveness.
4. Cultivate Listening & Empathy
Develop strong listening and empathy skills by adopting a growth and beginner’s mindset, taking time to absorb information, learn, and understand customers before formulating opinions or jumping to ideas.
5. Proactively Challenge Assumptions
As a product manager or leader, proactively challenge your own assumptions with an open mind; if there’s no conflict or contention, you might not be listening well or picking up on important cues.
6. Focus on Impact, Not Promotion
To get promoted, focus on creating significant business impact and doing what is right for the company, rather than optimizing actions solely for personal promotion; make your ambitions known to your manager.
7. Elevate MVP Bar for Safety
When safety is paramount, set an extremely high bar for the Minimum Viable Product (MVP) and do not cut corners, as core product management philosophy applies but with a different safety threshold.
8. Write a Press Release First
Before building a product, write a press release for the finished product to thoroughly think through and articulate its value proposition for users.
9. Demonstrate Commercial Progress
To maintain investor confidence and buy-in for long-term projects, demonstrate meaningful progress not just in technology, but crucially in commercial deployments and real-world results.
10. Focus on Customer & Business Value
Focus on creating genuine value for customers and building a sustainable business that makes sense, rather than optimizing for short-term investor brownie points.
11. Design for Credibility & Trust
Integrate credibility, predictability, and trust into the core design philosophy from the very beginning, ensuring users feel comfortable and safe with the system.
12. Build Trust via Transparency
Develop user trust by providing transparency into system actions (e.g., car’s view on a monitor) and ensuring human support is available for issues (e.g., calls for seatbelt).
13. Adapt to Human Expectations
Modify system behavior to align with subconscious human expectations and natural experiences, even if not strictly necessary for safety, to enhance user comfort and trust.
14. Mimic Good Human Behavior
When designing AI or automated systems, train models on human data, but filter out ‘bad’ data to mimic only good human behavior, making the system feel natural and trustworthy.
15. Design for Social Norms
Incorporate understanding of social norms into system behavior, as driving (and other interactions) are highly social and require adapting to unwritten rules in different contexts.
16. Balance Safety with Progress
Ensure products balance safety with adequate progress and assertiveness, as a system that is too cautious to move is not effective for users.
17. Track Comprehensive Metrics
Track both commercial/operational metrics (e.g., trips, active users, cost) and system behavior metrics (e.g., safety, rule compliance, progress) to measure product performance and business health.
18. PMs: Go Technically Deep
Product managers, especially in complex technical fields, must be able to go technically deep into the details of the product.
19. Embrace Uncertainty & Long Game
Be okay with uncertainty and ambiguity, and have the tenacity to play the long game, continuously improving the product, especially for transformative, long-term projects.
20. Prioritize Mission-Driven Work
Be driven by the mission of the product, such as making roads safer, as this passion is crucial for tackling challenging, transformational problems.
21. Adhere to Rules for Trust
Design systems to adhere strictly to rules, like speed limits, as users appreciate and trust predictable, rule-abiding behavior, even if it means sometimes going slower than human drivers might.
22. Understand Stakeholder Constraints
To improve listening and empathy, repeatedly practice understanding the constraints, perspectives, and definition of impact for different stakeholders in various environments and cultures.
23. Seek Challenging Projects
Improve your skillset and seek out challenging, high-visibility projects that stretch your abilities, dedicating yourself to creating significant business impact for the company.
24. Implement ‘Rule of Seven’ Emails
If an email thread reaches seven (or X) emails without resolving an issue, stop emailing and instead call the person or huddle in a room to resolve it directly.
25. Bring Your Own Playlist
For an awesome Waymo experience, bring your favorite playlist to play in the car and enjoy the ride.
5 Key Quotes
If there's no conflict, if there's no contention, then something is missing.
Shweta Srivastava
You need to disrupt yourself before somebody else does, because it's going to happen. It's inevitable.
Shweta Srivastava
The way to get promoted is to not want it too badly.
Shweta Srivastava
You have to be driven by the mission to make the road safer.
Shweta Srivastava
Bring your favorite playlist and back and enjoy the ride.
Shweta Srivastava
1 Protocols
Rule of Seven (or Ten) for Email Resolution
Shweta Srivastava- If an email thread reaches seven (or ten in larger companies like Waymo) emails without resolving the issue.
- Call the person or get in a room to huddle and resolve the issue directly.