宁可砍掉自己的项目，也不要押在单一增长渠道上

Host Lenny Rachitsky opens by noting how unusual Noam Lovinsky is for someone with his résumé: he barely tweets, has no newsletter, and has never been on a podcast. Lovinsky says he has wondered whether that is a mistake, but concludes that public posting simply does not come naturally to him. When he is working on something he goes deep and finds it hard to multitask beyond that.

Instead of networking at scale, he gets to know people by working with them and helping them, and he does a lot of coffees. He admits this does not scale the way Twitter does, but says it has served him well. Rachitsky adds his own view that many of the best people are too busy doing the work to post about it.

Lovinsky's advice to others is to do what you like, because you will generally be best at the things that energize you. He points to the number of technology waves he has lived through and argues that it makes little sense to spend your time on things that feel bad in the hope they lead to success. Leaning on what is authentic, he says, makes you likely to be among the best at it.

Lovinsky joined YouTube through the acquisition of a startup he had founded, initially rebuilding that product on Google infrastructure. Within a few months he looked at the wider roadmap and told leadership that putting 50 engineers on his project was not the best use of the team. He knew it might negotiate him out of a job, but felt it was right for the business.

Leadership agreed, wound the project down, folded some features into the existing product, and made him head of the creator focus area, one of three alongside viewers (led by Hunter Walk) and advertisers (led by Shishir Mehrotra). His takeaway is that in a healthy organization, advocating for what is best for the company will be rewarded; if it is not, you learn that early.

A second unusual move came when he struggled in the creator role while reporting to CEO Salar Kamangar, whose strategic questions felt to him like they came from another planet. He asked to report to Hunter Walk instead, reasoning that he would do better work with that support. It worked, and when Walk moved on he eventually shifted to leading the viewer PM team.

Asked where his confidence came from, he cites a broad view of company strategy and a nose for what a high-functioning team feels like, built from an early job among great people. He says that sense helps you choose teams, and trust that a good team will not push you into a corner for making the right call.

Rachitsky observes that a product leader's job is to work backwards from what is best for the business, not just the user or the team's goals. Lovinsky agrees but notes that customer and business interests usually overlap rather than always coincide. He recommends setting explicit principles for the moments when they conflict.

At Thumbtack the order was customers first, pros second, and Thumbtack last. Even in a marketplace where supply is critical, the team reasoned that what pros need most is more and better customers, so a great customer experience ultimately serves them. Raising prices to help the business while hurting marketplace liquidity would be locally optimizing rather than globally optimizing.

On when to give up on a project, Lovinsky says that what kills most early efforts is stamina. He recalls Thumbtack's CEO Marco describing the work as running uphill and chewing glass, and says resilience is worth practicing. But a team that has been hitting its head against the wall for too long may need a change of scene, and the real question is when you run out of steam, which usually happens before the money runs out.

Thumbtack was among the best companies at growing through SEO, but Lovinsky calls SEO a live-by-the-sword, die-by-the-sword channel, and says a one-channel company is always a no-no. Early in his tenure the company saw triple-digit growth, then a couple of SEO hits brought it to double digits, and eventually year-over-year growth turned negative for the first time. All of this happened while the team was rebuilding the product and changing the monetization model.

The recovery produced what a Sequoia board member, Brian Schreier, called the prettiest smile graph he had ever seen. For Lovinsky, the deeper lesson is that growth masks all problems. During the boom, Thumbtack was dropping demand on the floor because supply liquidity was insufficient, but with triple-digit growth the issue never felt urgent.

Once growth slowed and then reversed, the organization's tenor changed and people looked closely at what was really happening. He thinks it is healthy for businesses to go through such a moment, and that a PM who has only worked on growing products misses something. He now asks himself what he would do if his metrics went negative, which he says changes his sense of urgency and priority.

Lovinsky stresses that the turnaround was a team effort. First, Thumbtack turned on multiple growth channels, going back to first principles on paid channels and referrals it had tried and dropped, and assembled a stronger team around them. He also says the company was treating its targeting too narrowly: Thumbtack is really thousands of small marketplaces, such as DJs in Philadelphia or contractors in Sonoma, and growing market by market limited the data it gave to Google and others.

The core issue, though, was a high-friction experience. Customers answered a series of questions and were told they would hear back in 24 hours, while pros had to pay to quote on a job whether or not the customer had real intent. He praises that model as a perfect case of doing the scrappy thing that works, but says the business had outgrown it.

The team rebuilt the whole loop, changing the monetization model and enabling pros to give instant quotes across all those marketplaces. Rachitsky notes this mirrors Airbnb's shift from request-to-book to instant booking, which he led over three years. Lovinsky agrees that products involving both bits and atoms are exponentially harder, and that making such a change while still growing is deeply complicated.

Asked for other Thumbtack lessons, Lovinsky says the CEO's direct reports rarely get to do project work together, yet when growth goes negative their ability to tackle hard things as a group becomes critical. He concludes that no one can be a bystander on product strategy, whether CFO, head of people or head of engineering. Otherwise, hard times can degrade into a "what have you done for me lately" dynamic.

Rachitsky summarizes this as building relationships with the leadership team before things go wrong. Lovinsky agrees but adds an obligation on product and engineering leaders: bring strategy to the group in a form everyone can understand and engage with, so each leader feels their fingerprint is on it. Product managers are the connective tissue across functions, he says, but the whole leadership team should see itself that way.

Lovinsky joined the New Product Experimentation team early and helped build it, though he was not there when it closed. He pushes back on judging it by whether it produced the next Instagram, comparing that to asking whether a group won the lottery. His team was realistic that outcomes ranged from champagne-level hits to more modest but still beneficial results.

One benefit was rebuilding the muscle of starting small: at Facebook's scale it is impossible to run an experiment with a hundred people, and community products need density early on. Another was direct customer contact. Large companies often force PMs, engineers and designers to hear from users through recruiting agencies and reports because of legal and other concerns, and the incubator lifted those constraints and exposed what was broken for 0-to-1 work.

He also calls the team a fantastic recruiting tool, with many alumni going on to start companies, and notes it was well firewalled by its executive sponsor Schrep. Ideas like Threads passed through the team, and several ideas became features of other products because they could be validated faster without the mainline organization's constraints. Rachitsky adds an analogy to the Burning Man trash fence: some teams catch talented people who might otherwise leave.

Lovinsky's first piece of advice is to think hard about the incentive system, because smart people will game whatever you reward. A twice-a-year performance process is the wrong timeframe and creates adverse selection in who joins. It is hard to throw out company-wide leveling and pay processes for one group, but he argues it is necessary.

The second is letting teams choose their own infrastructure, knowing much of the code will be thrown away. At companies like Facebook or Google this takes a strong sponsor willing to say no to the usual rules. The third is organizational separation: he cites Nike's incubation lab, which recruits a different type of person under a different incentive system and brings in the distribution and marketing arms only after the team finds fit.

Rachitsky adds that the key competitor is the employee starting their own company, so upside matters. Lovinsky agrees and stresses time horizons: founders are not thinking about a promotion in six months but about outsized impact and outsized reward, and internal incubators should reflect that.

Lovinsky says Grammarly is much bigger by revenue than most people realize: it is 15 years old, profitable from day one, and still quite profitable. For years the founders deliberately avoided attention while others dismissed it as a feature that Google Docs and Word already had. With large language models, he says, the company now sits at the center of the hurricane, facing questions about ChatGPT and Copilot, though he still sees a bright future.

Having been there only about ten months, he offers his instinct on why users love it: how it works and where it works. You install it and it makes you better with no configuration, and assistance appears in the moment across every application and tab you write in. Delivering that quality at that scale, he says, is "a huge AI achievement masquerading as a little UX innovation".

The company's Ukrainian bootstrapped roots mean engineers think about how deep technical work turns into revenue within a couple of months. That culture helped Grammarly get here, but Lovinsky says it can also hold it back now that the law of large numbers applies. The business needs to invest ahead in new products, use cases and channels, and pull off a product-led shift from B2C to B2B, which its profitability now allows.

Both men reflect on the Ukraine team, many of whom have been with the company for over a decade and keep working through wartime disruptions. Lovinsky says the work can be a useful distraction and that the founders want Grammarly to succeed as a generational company in part for Ukraine.

Lovinsky says it is very hard to predict which job will succeed, but you can develop a nose for the people you can succeed with. He prioritizes positions that cause a lot of growth and learning, accepting that this can be painful and that the promised land lies on the other side. Rachitsky cites the line that "the cave you fear contains the treasure you seek".

Asked how to know when the pain is too much, he says every role should include one or two things where you can say "I got this", such as design sense, analytical thinking or deep familiarity with a product area. Those strengths form a foundation while you learn the rest, so be wary of situations that are entirely new. Rachitsky compares it to the flow-state chart where challenge is high but not overwhelming.

Lovinsky closes where he began: work on what makes you happy and gives you energy, because life is short and there is a lot to do. Rachitsky adds that people can run an energy audit to see what gives and drains energy, and do more of the former.