Lenny's PodcastTomer Cohen: How LinkedIn turned cringe feeds into signal
Through a two-million-member carved-out cohort and AI as matchmaker; clarity over confused beat hedging, and mountain-peak ambition revived the LinkedIn feed.
CHAPTERS
- 0:00 – 2:28
LinkedIn’s turnaround: the big bet behind making the feed worth checking
Lenny opens by calling out LinkedIn’s surprising transformation into a genuinely engaging daily habit. Tomer frames the turnaround by starting from LinkedIn’s full potential: an economic-opportunity platform built on a powerful social graph, where knowledge exchange can be a core transaction.
- •Why LinkedIn’s resurgence is “underappreciated” and what changed
- •Starting from the platform’s potential rather than current metrics
- •LinkedIn’s core identity: economic opportunity + social graph
- •Knowledge exchange as a high-leverage “economic transaction”
- 2:28 – 6:46
“We might be wrong, but we’re not confused”: a leadership mantra for alignment
Tomer explains the origins of his most famous phrase and why it spread culturally at LinkedIn. The core idea is that unified direction beats hedging, and clarity is what gives a team a real chance of success—even when the bet might be wrong.
- •Story of a founder using the mantra to escape organizational hedging
- •Why confusion in a system forces you to rely on luck
- •Separating being wrong (acceptable) from being unclear (dangerous)
- •How the mantra becomes part of an execution culture
- 6:46 – 9:17
Clarity of thought: define the real problem and take principled positions
Tomer digs into how to create clarity in product thinking: spend real time specifying the exact problem, audience, and nuance before jumping to solutions. He pushes for first-principles reasoning and strong opinions with explicit tradeoffs—principles that “have teeth.”
- •Product jams that focus deeply on the problem statement
- •From vague goals (“launch video”) to specific, scoped problems
- •First-principles solutions and forcing explicit tradeoffs
- •Distinguishing disagreement vs. misunderstanding to avoid wasted debate
- 9:17 – 13:10
Clarity of execution: priorities must match resourcing and talent allocation
Even when organizations decide, they often fail to act—creating hidden confusion. Tomer argues that priorities must show up in staffing, time, and top-talent deployment, otherwise the “real priority” is whatever teams are actually doing.
- •Common failure mode: decisions that don’t translate into action
- •Resource allocation as the truth serum for priorities
- •Why top talent shouldn’t drift to non-priority moonshots
- •How clarity and focus resolve execution drag
- 13:10 – 17:35
Ambitious goals + overdelivering: visualizing the peak (and the base camp)
Tomer shares another core belief: set truly ambitious goals to inspire and guide teams, then aim to exceed them. He uses a mountain metaphor—clearly see the peak and the starting point, even if the middle path is blurry.
- •Why “underpromise and overdeliver” isn’t inspiring for product building
- •The mountain metaphor: peak, base camp, and figuring out the middle
- •Ambition rooted in possibility, not constrained by current numbers
- •How ambitious peaks create alignment and energy
- 17:35 – 22:08
Rebuilding the LinkedIn feed: redefining purpose from promo to professional value
Tomer walks through why the feed needed a fundamental reset: it began as an activity tracker and drifted into promotional, cringey content. The turnaround started by forming a real feed team and redefining the feed’s purpose as professional knowledge exchange—people who matter discussing things you care about.
- •LinkedIn’s feed origins: activity feed → promotional drift
- •No unified feed team initially; creating ownership and focus
- •New purpose: not a traffic springboard or upsell feed
- •Target outcome: professional knowledge exchange and reputation-building
- 22:08 – 26:24
Running experiments at scale: carving out 2M members to escape internal politics
To change the feed’s ‘DNA,’ experiments created turbulence because many teams depended on feed traffic and metrics. Tomer solved this by isolating a cohort of two million members, giving the team freedom to iterate without destabilizing the broader org, and using results to prove the bet could grow the pie.
- •Why large-scale experiments triggered escalations and fear
- •The 2M-member carve-out as a protected sandbox
- •Behavior change evidence as the internal persuasion mechanism
- •Using negative tests and learning-driven experiments to validate direction
- 26:24 – 30:59
Metrics and marketplace health: measuring downstream, high-value engagement
Because the feed is the default landing surface, simple pageviews or sessions can be misleading. Tomer describes shifting measurement toward downstream, high-value engagement and balancing the marketplace: healthy creation and consumption, quality over raw volume.
- •Why top-level engagement metrics are noisy for a home feed
- •Defining ‘active engaged’ and ‘high-value’ downstream measures
- •Marketplace framing: creator-side and consumer-side health
- •Optimizing for the right views, not maximum views
- 30:59 – 36:13
AI-first as the feed’s engine: AI as ‘ultimate matchmaker’ for both sides
Tomer attributes much of the feed’s success to treating AI as the engine—not a supporting tool. The key is matchmaking at scale: creators reach relevant decision-makers, and consumers see content they can apply professionally, creating real economic opportunity and retention loops.
- •AI-first isn’t a feature; it’s the core mechanism for value exchange
- •Creator-side matchmaking: the right people seeing your expertise
- •Consumer-side relevance: content that’s actionable at work
- •Trading off ‘bad engagement’ and battling spam as objectives shift
- 36:13 – 40:57
How to build AI-first product teams: product leaders must own objectives, data, and infra
Tomer explains AI-first as a mindset that spans strategy, product, and talent—not a buzzword. He argues PMs must hold the ‘steering oars’: define algorithm objectives, choose features/parameters, invest in data collection and fine-tuning, and even engage at the infrastructure level for big leverage.
- •AI-first begins in strategy, flows into product, then hiring/talent
- •PM responsibility: write the objective function and understand the model levers
- •Data collection + fine-tuning as organization-wide strategy, not delegated work
- •Infrastructure and inference as product levers, not just engineering concerns
- 40:57 – 44:50
Operationalizing AI-first at LinkedIn: AI academy, reviews, and portfolio reset for LLMs
Tomer describes the internal mechanisms LinkedIn used to shift behavior: mandatory PM training via an AI academy, deeper leadership reviews on AI strategy, and building a bench of AI practitioners on the product side. After late 2022, they reworked product operations and portfolio focus to bring AI from the ‘back’ of matchmaking to the ‘front’ of user experiences.
- •AI academy training for PMs (similar to prior mobile-first transition)
- •Leadership review cadence that prioritizes AI strategy and objectives
- •Building internal product-side AI expertise to teach and scale practices
- •Portfolio and operating model changes to capitalize on the LLM wave
- 44:50 – 49:13
Letting go of roadmaps: principles for avoiding ‘AI everywhere’ low-value shipping
To prevent random AI add-ons, Tomer pushes teams to start from the same objectives and reimagine solutions with new capabilities—rather than starting from the tech. He also describes a deliberate pattern: allow divergence for exploration and learning, then converge via top-down focus on a small number of big bets.
- •Reset principle: ‘let go of your roadmap’ and return to problem clarity
- •Avoiding ‘cool tech’ traps by tying AI to existing objectives
- •Diverge then converge: exploration period followed by focused resource alignment
- •Prompt engineering as an internal playbook built through rapid learning cycles
- 49:13 – 56:52
Career growth lessons: conviction, craft, mentors, and choosing impact over status
Lenny zooms out on Tomer’s rapid rise at LinkedIn and asks for career advice. Tomer emphasizes loving the craft, learning from great people, and making career decisions based on conviction and impact—rather than demand signals, titles, or prestige—and he describes how that mindset shift changed his trajectory after moving to the U.S.
- •Learning from mentors (often informally) and absorbing patterns
- •A personal shift: from ‘what’s in demand’ to ‘what do I care about?’
- •Conviction as a prerequisite for strong product leadership
- •Product careers judged by shipped impact, not logos or titles
- 56:52 – 1:09:00
Takeaways + lightning round: books, Bluey, guitar tools, and LinkedIn video/jobs ‘coach’
Tomer shares final reflections, plugs his podcast ‘Building 1,’ and then moves through the lightning round (books, media, and favorite products). He ends with underrated LinkedIn features: the push into video for creators and an AI-powered jobs ‘coach’ experience aimed at reducing the loneliness of job hunting.
- •‘Building 1’ podcast: studying distinct ‘edges’ across builders and crafts
- •Top book recommendations: Mindset, Thinking Fast and Slow, High Output Management
- •Product inspiration from Bluey’s layered experience design
- •LinkedIn investments: video for creators and AI job-seeker coaching in the Jobs tab
