The Management Experiment Dividing Silicon Valley
Is big tech killing the one-on-one?
Written by Jennifer Guay | 7 min • July 28, 2025
The Management Experiment Dividing Silicon Valley
Is big tech killing the one-on-one?
Written by Jennifer Guay | 7 min • July 28, 2025
The word alone makes people flinch. Dr. Tessa West, a psychology professor at NYU, has watched it happen countless times in her research on workplace communication: mention “feedback” and you’ll practically see employees brace for impact.
Managers who avoid the dreaded word and instead weave feedback into daily tasks are getting something remarkable in return, West has found. They’re building stronger teams, catching problems early and creating the kind of authentic workplace relationships that most companies struggle to nurture.
As AI changes the pace of business and the very nature of work, some of Silicon Valley’s biggest names are rethinking traditional management practices. In May, Meta’s Mark Zuckerberg revealed he manages 25–30 people without any scheduled one-on-ones, arguing, “if you’re going to report to me, you need to be able to manage yourself.” The CEOs of Nvidia and Airbnb have also eliminated one-on-ones.
The shift away from formal one-on-ones reflects a deeper tension about how feedback should work in modern organizations. For decades, companies have treated feedback as something that happens during designated meetings: performance reviews, weekly check-ins and structured conversations. But what if the very act of scheduling feedback makes it less effective?
“Never use the word ‘feedback’,” West tells managers. “It's just scary.”
Her research reveals why traditional feedback systems often fail. In formal reviews, feedback can feel too personal — like an indictment of someone’s worth as an employee. Her work backs ambient feedback: brief, behavior-specific observations that happen in the moment, rather than during nerve-racking formal reviews. It may mean praising a clear graph in a presentation, for example, while noting that the y-axis needs better labeling.
“Performance reviews are dumb,” West said bluntly. “Feedback should take 10 seconds and be folded into the every day.”
West’s research also reveals stark generational differences in how people prefer to receive feedback. While all age groups want more direct feedback from managers, baby boomers value prescriptive feedback from leaders with insider knowledge and experience. Members of Generation Z, meanwhile, dislike top-down feedback — and may even find it offensive. They prefer empathetic, casual conversations with managers.
West recommends that managers first model receiving feedback from higher-ups to create a culture where these conversations feel normal, rather than punitive. Other management experts espouse a similar view: the best feedback happens when employees feel safe expressing concerns without fear of retaliation.
Gabe Larsen, who spent two years as a marketing director at Meta, got a preview of Zuckerberg’s thinking years before it made headlines. In 2022, after Zuckerberg announced 2023 as Meta’s “Year of Efficiency” during a Q4 earnings call, there was a “noticeable shift,” Larsen said.
“People started talking more about performance management and how we operate,” said Larsen, now an AI growth advisor at Signals. “That’s when it became clear his approach was different.”
It was eye-opening, Larsen said. For years, he had dutifully scheduled weekly check-ins with his reports, armed with agendas and talking points. “I’d never really thought about doing it differently,” he recalled.
But something about Zuckerberg’s reasoning clicked, and Larsen decided to run his own experiment. He eliminated scheduled one-on-ones with his direct reports in favor of what he calls “free-flowing” communication: impromptu feedback, ad-hoc strategy sessions and frequent phone calls. Over the past five years, this approach has helped him make decisions more quickly and build closer relationships with his direct reports, he said.
“The one-on-one structure hampers creativity and momentum,” Larsen asserted. “When you catch people a little off guard and they give you top-of-mind responses, it’s just so real.”
But not everyone is ready to abandon the one-on-one.
When engineer Jay Gengelbach first walked through Google’s doors in 2006, he was fresh out of college, eager and armed with a dim view of management. Weaned on “Dilbert” comics, he saw managers as “pointy-haired, incompetent idiots” who slowed progress with needless bureaucracy.
At the time, Google was still fine-tuning its organizational structure. Gengelbach said his managers oversaw between 40 and 75 direct reports, leaving it “up to the engineers to self-organize.” He enjoyed the freewheeling autonomy, but noticed himself gravitating toward incremental improvements over bold innovations.
As Google developed a more structured management program, Gengelbach saw what he’d been missing. Engineering wasn’t just about writing code — it was also about understanding how that code fit into the broader picture. “Whether it’s one-on-ones or team syncs or retrospectives, you need to appreciate how what you’re doing contributes to the overall success of the company,” said Gengelbach, who worked at Google for 12 years.
After all, research still suggests that regular one-on-ones work when managers use them to give real feedback, offer support and allow employees the space to voice concerns. Last year, McKinsey concluded that companies with thoughtful performance management processes see 30% higher revenue growth than competitors.
"Last year, McKinsey concluded that companies with thoughtful performance management processes see 30% higher revenue growth than competitors. "
Now an engineering leader at Vercel, Gengelbach applies the lessons he learned at Google. He uses one-on-ones not as perfunctory status updates, but as conversations about fulfillment, obstacles and growth: “Do you feel like you're being your best self this week? What’s holding you back from doing the things you want to be doing in this company?”
Those, he said, are “the good one-on-ones” — the kind that create space for real feedback and dialogue, not just a mechanical exchange of information. These deeper conversations don’t always happen organically during the daily rhythms of work, he believes. Some structure helps ensure they don’t fall through the cracks.
West’s research theoretically backs Zuckerberg’s management philosophy: with feedback baked into the every day, there’s little need for formal one-on-ones. She has one important caveat: feedback must still happen consistently. Too often, managers avoid direct feedback altogether, defaulting to generic positive comments that help no one.
Ineffective feedback processes can also cost companies, West warned. She points to NASA and Lockheed Martin’s failed mission to Mars, when a simple miscommunication about units of measurement precipitated the loss of a $125 million spacecraft. “A lack of individual-level feedback shows up as team failure, which tends to be a very expensive problem,” she said.
Even Larsen admits that holistic feedback and career development can fall by the wayside without time set aside for the broader conversations Gengelbach described. This approach may disadvantage employees from minority groups who lack the social capital or insider knowledge to navigate advancement — precisely the people that structured feedback systems were designed to protect.
For an industry that prides itself on rapid iteration and continuous improvement in product development, the irony is that many tech companies have failed to apply the same principles to employee development. West’s method treats feedback like a code review: it’s frequent, specific and focused on improvement, without judgment.
Ultimately, the format matters less than the content. Whether it’s delivered in a structured weekly meeting or over a quick coffee break, employees need feedback to guide their development. As AI becomes more embedded in work culture, the distinctly human skills that define good managers — strategic thinking, creative collaboration and complex problem-solving — will only become more critical. These capabilities don’t develop in isolation. And they may soon prove essential to managing a new kind of direct report: AI itself.