像产品负责人那样看清 Matrix：Scott Belsky 让人爱上产品的几条铁律

At Lenny & Friends Summit, Scott Belsky opens by outlining a plan: share products he has been exposed to, a curated set of product insights, and a closing thread about the human side of product work. He describes his long-standing passion as organizing the creative world at work, which began as a physical product and grew into books, conferences and Behance, a platform of 56 million creatives. Behance let people sort work by person, agency, tool, discipline and color.

Scott explains that Behance joined Adobe in 2012; he expected to stay a few years but left after three and returned for another seven, overseeing Creative Cloud and bringing products such as Lightroom mobile and Adobe Express to web and mobile. He also works with startups as an early product adviser and seed investor and writes about product. He then asks what it means to see the Matrix as a product leader—moments when something clicks into a cardinal insight or lens that changes decisions.

Scott's first cardinal insight is that data is a compass, not a map. Intuition, understanding and empathy determine which mountain to climb; data helps the team climb the mountain it has chosen. In a discipline increasingly driven by data, he says, teams must remember that distinction.

For new products, Scott urges teams to prioritize the problems they want to have. Success means customers use a product enough to request things power users would want; failure means they get stuck in onboarding or blocked by security. He frames roadmaps on a timescale where shallow value must be optimized before deep value, because if any product gets more than 30% of the top of the funnel through, it is world-class.

That logic favors quick feelings of success—product-led growth's definition in part—before deeper value. Scott also warns that an MVP has gravity: teams instinctively iterate from whatever they ship. The differentiator must therefore be done extraordinarily well before launch, because it is hard to iterate out of missing that piece.

Scott calls the best product practice a bonsai discipline: do only half of what you want to do and prune beautiful branches to enrich the core. At Behance, the team had groups, Tip Exchange, circles and work in progress; each time they killed one, engagement on the core act of publishing projects to a portfolio rose.

He describes a product life cycle in which users flock to simple products, the product takes them for granted by adding power-user features, complexity grows, and users flock to simplicity again. Keeping products simple requires serving the newest cohort, because early customers are more forgiving while later cohorts demand a simple, different onboarding. Scott argues that customers in early stages—even enterprise customers—are lazy, vain and selfish.

The response to laziness is to show rather than explain: sliders and tooltips in Photoshop demonstrate actions in context. Scott says the only thing better than showing is doing the work for the customer, which is why templates, chat UIs and presumptuous defaults matter. Adobe tried building new features on top of features so the product does things for customers instead of teaching them how.

Scott's favorite saying, borrowed from Dave Marin, is that the devil is in the defaults. Teams often think about default state in the final mile of shipping, but it may be the only state most customers ever experience. Arguing about that default experience is therefore crucial.

Customers are also vain: products must make them look good quickly. Scott calls this ego analytics—how a product makes someone feel good about themselves, their job or their life. He cites social products' hearts, likes and appreciate buttons, and notes that people who work at Instagram have said engagement rises after a user posts, suggesting users return to see who saw their content.

Behance let everyone appreciate work even when not logged in because the gratification of being appreciated mattered more than restricting feedback. Scott also invokes Harvard Business School research on motivation: people make more progress when they feel they are making progress, so leaders should merchandise team progress and products should show customers their own progress. Finally, selfishness means customers must benefit quickly from a little time, as with Pinterest's early bookmark utility, Square's faster payments and Stripe's two lines of code.

Scott urges product leaders to get customers talking about their problems rather than the product. In early Behance focus groups, people said they did not need another social network because they already had DeviantArt, Facebook and LinkedIn. When asked what they suffered from, they revealed that they lacked attribution for their work, their portfolios were out of date and they were not reaching potential employers.

That leads to his claim that empathy for customer problems outperforms passion for solutions. Among the seed investments he was most inspired by but that did not succeed, founders were governed by passion for an end state and lacked empathy for what customers experienced daily—even when customers said they wanted the product. That gap, he suggests, can leave a company 30 degrees off product-market fit.

Scott also believes the most powerful and affordable driver of growth may not be what customers say they want. HipChat was the team's tool before Slack, but Slack spread because of animated GIFs and random sends to colleagues; then people used it to ask who wanted coffee, and it became a daily utility. The lesson is that novelty can precede utility, so roadmaps should leave room for unexpected delight.

People do not rave about products doing what they expected—they rave about what they did not expect. Scott points to Tesla owners talking about the Rainbow Road easter egg rather than the car functioning as a car. He acknowledges the difficulty of prioritizing Rainbow Road in a roadmap session, but argues that adoption depends on such moments, so leaders must prioritize them.

Whenever Scott reviews a screen, he asks three questions: How did I get here? What do I do now? Where do I go next? He calls this the ultimate object-model test, because customers are busy and distracted, and a product that fails these questions is like dropping someone in the desert without a map.

The questions force clarity about actions, emphasis and defaults: are the next actions clear or confusing, and is the product pushing enough emphasis in the most likely direction? Scott treats these as a Holy Grail for product reviews. They turn a design review into a test of whether the customer can stay oriented.

Scott calls A+ designers a cheat code and says his proudest career moments came from partnerships with designers. A prototype is worth 100 meetings: when a team debates whether products can connect or a feature makes sense, a designer's prototype cuts through and prompts decisiveness.

Perceived performance can matter as much as actual performance. When Pinterest scaled in 2011 and 2012, designers suggested loading the color of a pin before the pin itself; customers immediately said Pinterest felt faster even though nothing under the hood had changed. In the AI era, Scott adds, more choices are made at the surface, so design may increasingly commoditize technology and designers need a seat at the table.

Scott says leaders must prioritize grafting talent as much as hiring it. Hiring an amazing person is not enough if the team's immune system rejects the new organ; leaders must suppress the immune system so a senior person who does things differently can take hold and change how the organization works. The industry invests heavily in hiring, he notes, but treats grafting as if a computer and login will suffice.

On hiring, Scott favors initiative over experience. People with a history of taking initiative can exceed expectations when given the chance, especially when a company cannot afford the most experienced candidates. He also argues for celebrating impact over tenure: at Adobe he replaced quarterly tenure awards with impact awards so new contributors felt valued regardless of how long they had been there.

He urges leaders to collapse the talent stack whenever possible—hiring the product leader with a design background, the designer who can engineer, or the growth hacker who also designs. That conduit unlocks something new, and Scott says it has always surpassed his expectations. He frames this as an opportunity to create an incredibly effective conduit inside the organization.

Scott also insists that you can only expect what you inspect. Each product leader must find the few level ones they hold themselves directly accountable for and inspect rigorously, even while delegating almost everything else. He learned this from a CEO who delegated broadly but went deep on a short list of inspections.

Scott advises product leaders to know when to amplify their own intuition amid data, best practices and outside advice. Conviction should outweigh consensus when someone you trust feels strongly, because that is how non-popular, differentiating decisions get made. When frustrated or disappointed, he tries to lead with curiosity over criticism by asking a barrage of questions until the team—or he—realizes what is wrong.

As product management offloads more work to compute and agents complement managers, Scott expects the job to shift from how to do something toward what to do. That makes taste more important than skill as skill is offloaded, and the what-to-do remains a human discipline. Models are trained on the middle of the data distribution, he says, while great decisions and creative solutions fall outside what has been done before—otherwise they would already have been done.

His litmus test for new products is that the best ones take people back to the way things once were, but with scale and efficiency. People long to be known in their towns, so marketing technology at its best helps them be known again; LinkedIn's networking and referrals recall the old way of getting jobs rather than anonymous job boards; AI interfaces return people to natural language.

Scott closes with the thread that product is ultimately a people discipline. Product decisions must boil down to deep understanding and empathy for human tendencies—lazy, vain and selfish, but also seeking that first smile—and to understanding the parts of the past people long for, with scale and efficiency. He adds that a product reveals who did and did not talk to each other: whether the copywriter was in the room, whether marketing needed product sign-off before shipping. The Apple does not fall far from the tree.