How LinkedIn became interesting: The inside story | Tomer Cohen (CPO at LinkedIn)
Tomer Cohen, Chief Product Officer at LinkedIn, shares insights on transforming the LinkedIn feed, fostering an AI-first mindset, and leveraging his "we might be wrong, but not confused" mantra to drive clarity, execution, and ambitious product leadership.
Deep Dive Analysis
12 Topic Outline
Introduction to Tomer Cohen and his product leadership philosophy
The 'We might be wrong, but we are not confused' mantra
Importance of clarity of thought and execution in product building
Setting ambitious goals and aiming for significant impact
Strategy behind transforming the LinkedIn feed into an engaging platform
Running large-scale experiments and carving out user cohorts
LinkedIn's evolution to an AI-first mindset and its implementation
Shifting product development from deterministic control to ingredient management
LinkedIn's approach to integrating new AI waves and revisiting roadmaps
Tomer Cohen's career growth and personal philosophy
The 'Becoming is better than being' life motto
Leveraging video content and the AI-powered job seeking coach on LinkedIn
5 Key Concepts
"We might be wrong, but we are not confused" mantra
This phrase emphasizes clarity and principles in leadership, meaning that while a chosen path might ultimately prove incorrect, the team is unified and clear on the direction and rationale, avoiding internal hedging and confusion. It prioritizes collective forward movement over individual attachment to being 'right.'
Minus one to one products
This refers to turnaround products or initiatives that start from a negative perception or underperformance and aim to become successful. The challenge is often harder internally due to entrenched flows, processes, and metrics that need to be fundamentally shifted.
AI-first mindset
This is a strategic approach where AI is considered the core engine of a product or organization, not just a technology. It involves product leaders understanding and directing AI's objectives, features, data collection, and fine-tuning, rather than delegating it as a black box.
Product leader as a chef
This analogy describes the shift in product leadership with AI. Instead of dictating every aspect of the user experience (like a chef controlling every detail of a dish), the product leader focuses on providing the best 'ingredients' and 'guidelines,' allowing the AI to dynamically create the optimal experience.
Becoming is better than being
This life motto emphasizes continuous growth, learning, and evolution over reaching a fixed state of achievement. It suggests that the moment one believes they have mastered something, they risk becoming obsolete, highlighting the importance of a growth mindset.
6 Questions Answered
It means that while the chosen direction might not be perfect, the team is completely clear on the problem, the solution's principles, and the execution plan, ensuring everyone pulls in the same direction to maximize the chance of success.
The strategy involved setting a new purpose for the feed focused on professional knowledge exchange, making it AI-first by unifying the AI team with product objectives, and carving out a 2 million-member cohort to experiment and prove the new experience's value before rolling it out broadly.
Product leaders should treat AI as the core engine of their product, understand the algorithm's objectives and features, invest in data collection and fine-tuning, and encourage teams to revisit existing problems with new AI capabilities, rather than just finding uses for new tech.
A key factor is pursuing what one has strong conviction and passion for, rather than solely focusing on what is in demand or challenging. This involves identifying areas where one can make a significant impact and continuously learning and evolving.
LinkedIn's approach involved first allowing teams to explore and inspire creativity with AI by revisiting existing problem statements, then converging resources on the most promising bets, and providing top-down guidance to ensure focus and quality.
LinkedIn offers an AI-powered 'coach' experience within the jobs tab, which provides personalized, private support for job seekers, helping them brainstorm, compare opportunities, and feel supported throughout their often lonely job search journey.
14 Actionable Insights
1. Embrace Clarity & Alignment
Adopt the mantra ‘we might be wrong, but we’re not confused’ to ensure everyone is clear on objectives and aligned on execution, which gives a team the best chance of success and fosters learning from decisions.
2. Adopt an AI-First Mindset
As a product leader, take ownership of AI strategy by defining algorithm objectives, identifying features to learn on, and investing in data collection and fine-tuning, recognizing AI as the primary engine of product success.
3. Start with Potential, Work Backwards
Begin by envisioning the ultimate potential of a product or initiative, rather than current limitations, to set ambitious and inspiring goals that drive significant impact and growth.
4. Isolate Experimental Cohorts
When transforming an existing product in a large organization, carve out a small, randomized user cohort to experiment freely and prove new experiences without disrupting overall metrics or internal politics.
5. Define Nuanced Problem Statements
Spend significant time defining the exact, nuanced problem you’re trying to solve, including target audience and unique criteria, to clearly visualize the path forward and ensure focused solutions.
6. Develop Principles with ‘Teeth’
Formulate product principles that include a willingness to sacrifice or trade off certain vectors, as this demonstrates strong opinions and leads to more impactful and focused decisions.
7. Clarify Disagreement vs. Misunderstanding
When encountering resistance or unclear communication, push to determine if someone is genuinely misunderstanding a point or actively disagreeing, to resolve issues more efficiently and avoid wasted time.
8. Align Resources with Priorities
Ensure that stated top priorities are genuinely reflected in resource allocation, including engineering talent, to manifest decisions into effective execution and avoid confusion about what truly matters.
9. Set Ambitious, Inspiring Goals
Visualize and communicate the ‘peak’ or massive impact you aim to achieve, even if the middle path is blurry, to inspire teams and drive significant change, rather than underplaying potential.
10. Pursue Passion and Conviction
Dictate your career path by what you care about, are excited by, have strong conviction in, and where you can make a significant impact, rather than solely by external demand or challenge.
11. Learn from Great People
Actively seek to absorb knowledge and insights from mentors and managers, even informally, to continuously grow and improve your craft throughout your career.
12. Embrace ‘Becoming, Not Being’
Adopt a growth mindset by focusing on continuous learning, evolution, and skill development, recognizing that mastery is an ongoing journey rather than a fixed state or destination.
13. Leverage LinkedIn Video for Creators
Utilize LinkedIn’s immersive video features to share expertise, as the platform’s AI prioritizes getting content to the right professional audience, leading to significant opportunities and influence.
14. Utilize LinkedIn’s AI Job Coach
For job seekers, use the AI-powered coach in the jobs tab to get personalized, private support for fit, application strategies, and career brainstorming, helping to alleviate the loneliness of the job search.
7 Key Quotes
We might be wrong, but we are not confused.
Tomer Cohen
I don't mind being wrong, it really comes from a humble place. I would rather go forward with everybody in the same direction than necessarily try to hedge all the time, which will give me no chance of success.
Tomer Cohen
AI is the ultimate matchmaker. It's underutilized. It's misunderstood.
Tomer Cohen
If you're not sure about what we're trying to accomplish, how can you know what you learn from it?
Tomer Cohen
I don't get attached to what have, did not work in the past. That's not, I don't know, maybe it's a mistake sometimes, but I don't, that doesn't stop me from thinking about the future.
Tomer Cohen
I think every technological revolution has dramatically changed the way we build. And AI, arguably, is the biggest one in our lifetimes.
Tomer Cohen
Becoming is better than being.
Tomer Cohen
2 Protocols
Transforming an 'entrenched' product (Minus one to one product turnaround)
Tomer Cohen- Set a new, clear purpose: Define what the product should be, not what it currently is (e.g., LinkedIn feed as a knowledge exchange, not a promotional springboard).
- Make it AI-first: Integrate the AI team directly with product objectives, recognizing AI as the core engine.
- Unify AI objectives: Ensure the AI team's goals align with the new product purpose.
- Carve out a user cohort: Isolate a segment of users (e.g., 2 million members) to experiment with the new experience, allowing for full liberty in development without impacting overall metrics or causing internal escalations.
- Gather strong evidence: Observe dramatic behavior changes and positive outcomes within the carved-out cohort over months.
- Scale the proven experience: Once validated, integrate the new experience into the main product for all users, focusing on growing the overall value rather than just re-allocating existing 'pie.'
- Focus on professional opportunities: Ensure the platform facilitates valuable interactions, like connecting creators with relevant audiences, and actively remove bad engagement.
Developing an AI-First Mindset in Product Teams
Tomer Cohen- Establish an AI academy: Provide mandatory training for all product managers to understand AI fundamentals and its role in product development.
- Focus reviews on AI strategy: Incorporate AI strategy and algorithm objectives into product reviews to ensure leadership alignment and understanding.
- Hire AI practitioners: Bring in strong AI product leaders and experts who can teach and guide teams.
- Revisit roadmaps with AI in mind: Encourage teams to let go of existing roadmaps and re-evaluate problem statements, asking how AI can solve objectives better.
- Allow for creative exploration (Diverge): Give teams time and space to experiment and build different AI-powered solutions, even if it leads to some duplication initially, to foster learning and excitement.
- Provide top-down guidance and converge: After an exploration period, identify the biggest bets and converge resources on those, providing clear focus and ensuring efficient use of capacity and cost.
- Shift control to ingredients: Recognize that with AI, product leaders control the 'ingredients' (data, guidelines) rather than dictating the exact user experience, embracing AI's non-deterministic nature.