Seth Godin：用品味、张力和四个选择，做出值得被人谈论的产品

Seth defines good taste as knowing what other people want just before they do. He says a private preference for peanut-butter-covered licorice may be legitimate, but it is not good taste because everyone else finds it weird. People with good taste—whether jazz musicians or fabric designers—bring something the world did not expect but is glad to see.

Seth separates quality from luxury and perfection. Quality means meeting spec, and if you meet spec you are done; if the spec is not good enough, make a better spec. High standards mean relentlessly improving the spec in service of the people you work with, not refusing to ship because you are a perfectionist—that is hiding. At Spinnaker, he learned the standard was to delight the software buyer, not to please a boss with excuses.

Seth extends the point to any job: there is no perfect job, but you can choose to be the best barista for an hour rather than be mediocre and bored. He cites Martin Luther King Jr. and a jazz saxophonist friend who still has to schlep between half-empty rooms. The boundary of the work is real, but within it people can add delight for their own sake, not only for pay.

Seth got his first real product-management experience in a summer job at Spinnaker Software, the 30th employee at a company that invented educational computer games. He did assistant brand-manager work—meetings with the ad agency, copywriting—but at the end of the summer the president, chairman, and head of marketing pulled him into a secret project. They wanted illustrated, music-filled computer adventure games based on science-fiction novels and had acquired work by Ray Bradbury and Arthur C. Clarke.

Seth had read every science-fiction book in the Clearfield Public Library, so the project was a dream, but his assigned job was packaging and ads. He discovered that engineers had many people competing for their time, and without a way to get more than his fair share of development he would have no products—and no job. He started an internal newsletter using desktop publishing, mentioned engineers who worked on his project, and put printed copies in every office mailbox.

Within three or four months, partly because the project was good, about 40 engineers were working for him even though no one reported to him; he became the de facto product manager. That is where he learned that marketing is the product—you do not make a product and hand it to a marketing person to put a logo on it. Five products in a row went gold, and near the end he stayed in the office for 22 days, sleeping four hours a night, because RadioShack had bought a huge order and a missed date could bankrupt the company.

The product shipped, and the next day Herbie Hancock called to say he had stayed up all night playing Fahrenheit 451 and loved it. Seth recalls the thrill of making decisions, cajoling a team that was not in it for the money, and shipping something they were proud of. The story is his origin as a product manager and as someone who sees marketing and product as inseparable.

Seth offers three product lessons off the top of his head. First, empathy is not about kindness and it is not optional: the product is being made for other people, so telling customers to read the manual is a maker's mistake, not a customer's mistake. Second, when a project runs out of time and money, it is over, so professionals do not ask for extensions just because they had good intentions—unexpected things are part of the deal.

Third, especially for software as a service, if you do not build a network effect into what you are making, you are almost certainly going to fail. Seth says you cannot hope Apple will pick you, promote you, and make you successful; the question is whether the product works better for users if they tell other people about it. If the answer is no, why would they tell anyone, and if they do not tell anyone, no one will hear about it; if the answer is yes, the marketing problem is largely solved.

Asked how AI companies can differentiate, Seth starts with what a brand is: not a logo but a promise, what someone expects from you and whether they would miss you if you were gone. Hyatt has a logo, Nike has a brand—if Nike opened a hotel people would know what it would be like, but if Hyatt made sneakers people would have no clue. To build a brand, a company must stand for something and say what it does not do.

Seth expects AI to stop being a feature the way electricity is not a feature: many things run on electricity but are not electricity companies. AI companies and all companies therefore need to answer what is in it for the user, make a difficult and remarkable promise, and then keep it. Absurd promises might work at the VC level but not with consumers, because you can only break that promise once.

Seth uses airlines to illustrate the difference between loyalty and bribery: people stick with an airline for points, but loyalty would mean paying extra to stay. He says he has an emotional connection to Claude.ai because it has a brand, while ChatGPT regularly overpromises, underdelivers, and lacks kindness or humility in his experience. Claude, by contrast, is clear when it does not know and is not lazy or angry, so Seth happily pays extra for it.

Seth also used Claude as a writing assistant for This is Strategy, while insisting that every word with his name on it is his. He would upload a list of four things and ask what he missed, or upload chapters and ask which claims he was making but not sustaining. Claude became a patient editor—something hard to find in the real world because editors are not paid enough or given enough time—and Seth used it to diagnose when a sentence did not sound like him.

Lenny introduces Seth's book This is Strategy and the four-part sequence that stuck with him. Seth says these are critical choices for product people, who often gloss over them as givens and sacrifice agency over the four most important things they should choose.

First, choose your customers: if your motto is that anyone can be a customer, you will not succeed, but a smallest viable audience defines language, budget, problem, technical facility, temperament, and retention. Seth points to the Humane Pin as a company that picked customers who expected a finished Apple-quality product on pre-order, then shipped something that was not that; by contrast, launching to 400 Raspberry Pi-like wonks on Product Hunt would have let the company dance with its actual constraints. A wedding photographer in the Hamptons should expect spoiled, cranky clients because the customers were chosen.

Second, choose your competition: competing with Walmart means accepting that Walmart lowers prices, and that choice builds boundaries around what you do. Third, choose your source of validation: if you try to please your boss, that is what you will do, but one difficult conversation can establish that you are matching the taste of the 400 agreed customers rather than the boss's taste. Fourth, choose your distribution, which matters enormously in software; when Seth's video-game company was in Target and Lechmere while competitors were in Ziploc bags in computer stores, the arrival of Steam or downloadable shareware changed everything about the product.

Seth stresses that these choices are intertwined and not easily changed at will, but product people must build with them in mind. A friend's food business plan that proposed Shopify plus airports, hotels, and businesses drew Seth's objection because a two-person team cannot serve all those customers and channels. The discipline is to decide today who you are for and who you are not for.

Seth distinguishes tension from stress. Stress is two conflicting impulses at once—wanting to leave but needing to stay, hating a boss but needing the job—and it is not a good thing to seek. Tension, by contrast, is at the heart of every art form and every innovation: it is the possibility that it might not work, the unfinished musical phrase the listener needs to complete.

When a product launches, the maker says it can do X, and the customer imagines what life would be like if that were true. If they fall in love with that possibility, tension appears: did you tell the truth, and will it work for them? Seth remembers PF Flyers sneakers before Nike, whose promise was to make a kid fast enough to run away from the bully; at ten years old in the shoe store he felt real tension because he needed to get away and was holding the company to its promise. The company did not tell the truth, which connects back to the brand promise.

Lenny brings up Purple Cow, and Seth starts with Steve Blank's customer traction: the best sign a startup has hope is whether engaged users stick around, come back, and tell their friends—and what they say. Remarkable means worth making a remark about, not a gimmicky viral video. If a product makes a user's life better when they talk about it, and the maker knows in advance what they want users to say, users are more likely to say it.

Seth credits Marissa Mayer with creating enormous stock-market value by refusing to add more buttons to Google's homepage and keeping it to two buttons. Yahoo, where Seth worked, had 183 links on its homepage while Google had two; Google's design promised to be smart enough to take users where they wanted to go without browsing. Seth, a tech-forward person, would send lost friends to Google because he knew they would not come back for more help—and telling them raised his status as a tech innovator.

The implication is that most companies win through word of mouth, not paid ads or sales teams (unless they are selling B2B software), so they must build something people tell friends about. Seth says this is not only for small companies: Microsoft beat WordPerfect because Word became the format coworkers needed on a modern computer, and once people migrated they did not go back. It was in each user's interest to insist that others use Word, so they did.

Seth says systems are everywhere and largely invisible, and they protect themselves by inventing culture—the way things are around here. High-school seniors do not see the college industrial complex because growing up in the suburbs makes it seem normal. If something feels scary, that may be the system telling you not to test it; the response is not to be fearless or foolhardy but to ask why it scares you, what system would be offended, and whether you are working with the system or leveraged enough to change it. He illustrates with downloadable software: the first people to try it were told it would be pirated and would never make money, but Seth could see why the dominant distribution system did not want him to experiment. Since he had only worked on the software for five months, he could try it and make new software if it failed, and Salesforce later made 'no software' and the cloud the new system, which turned the old certainty into the risky move.

Systems enable interoperability—right-hand handshakes, the metric system, zip codes—but they can start doing things to support themselves. Seth points to job interviews: the most successful interviews often reward candidates who look, went to school, and have backgrounds like the interviewers, which is toxic because it reduces both diversity and access to real talent. If a company's last four hires were Ivy League tennis captains, there is a pattern being reinforced by people who do not want trouble, and the system is not helping them. Naming that system is the first step to changing the rules.

Seth uses the line 'better waves make better surfers' to argue that success depends heavily on choosing the wave, not only on skill; his son passes up waves others would take because he knows he can surf the right one better. Choosing which company to join, which people to put in key roles, and whether to wait for a superstar rather than ship mediocre work to meet an imaginary deadline are all wave choices. Mediocre means average, so if you do what others do, you should expect what others get. On Jaguar's rebrand, Seth says Jaguar did not rebrand—it re-logoed, and walking away from an iconic, beloved brand with almost no cars that undermine what it could stand for is a significant error made to please an art director. He compares it to IHOP's fake name-change announcement and Oreo's Super Bowl tweet: buzz does not sell hamburgers or cookies, and Jaguar will not sell more cars because people are talking about its re-logoing.

What sells cars is customer traction. Seth says Tesla's early Model S did this well: people talked about the pop-out door handles in supermarket parking lots, then the owner would invite them in and hit ludicrous mode, and the passenger would scream. That was fun for the owner and told a story—a form of customer traction you could not get in another car. By contrast, the Cybertruck was intentionally divisive, and divisiveness does not sell to pickup-truck buyers, a category where Ford's F-150 can represent more than 100% of Ford's profit; Tesla could have crossed the chasm by selling utility to skeptical buyers instead of pulling a stunt. Lenny closes with Seth's leadership line—do what you believe, paint a picture of the future, go there, and people will follow—and Seth refines it: the picture must be based in empathy, showing where they want to go, not where you want to go, because professionals are not entitled to money, market share, or a good price, only to be of service.