好名字出现时你几乎认不出来：Lexicon 如何给 Sonos 和 Azure 起名

Placek's favorite story is Sonos. The founders were stuck on a name that placed them in the entertainment business, and they rejected Sonos because it lacked emotion. He pushed back, arguing that they made speakers that let entertainment flow, and that Sonos was about sound. The name is also a palindrome, which he says helped get them thinking.

When the founders were still struggling, Placek flew back to Santa Barbara, without billing for the trip, to make the case again. Founder Bob MacFarlane then said they had been naming the company for themselves when they should be naming it for the market and its customers. That shift is what turned Sonos into the right name.

Placek says this pattern is the norm across nearly 4,000 projects. Most clients arrive confident that they will know the right name when they see it, but he says that almost never happens. Humans seek comfort, and one source of comfort is something that has succeeded before, which is why even Harry Potter was reportedly turned down many times. His answer is to treat naming as creating the future, not copying the past.

He also says Lexicon starts conversations by asking how a company behaves now and how it wants to behave in future, not about positioning, values or mission. That kind of questioning, he argues, was fine 25 or 30 years ago but does not produce names like Sonos in a complex, digital world.

Microsoft first came to Lexicon asking for a name that started or ended with "cloud." Placek warned that this would drown the product in an ocean of cloud-this and cloud-that. The team did produce cloud names on request, but also presented Azure, a word for blue that links to sky and clouds and has a noisy Z followed by a smooth ending.

One client called Azure "a dumb idea." Placek says that after four decades such remarks roll off him, and that the client team warmed to it as they discussed it. He told them they did not want to make a statement, they wanted to start a story, and he notes that Azure is now a brand worth something like $100 billion.

He argues for two kinds of advantage. Cumulative advantage builds as customers see a distinctive name more often and bond with it, since nothing is used longer than a name. Asymmetric advantage means starting ahead of the market on day one, which a descriptive name like Cloud Pro cannot give you among ten similar services.

Placek illustrates asymmetric advantage with the Melian Dialogues, in which the Athenians arrived at Melos having already secured an overwhelming advantage. He calls it one of the best books on marketing, even though it never mentions marketing. Lexicon's mission, he says, is not a good name but the right name.

Vercel CEO Guillermo Rauch told Lenny that before working with Placek, naming felt like charisma, something you either have or don't, and that watching the team turn it into a science was surreal. Placek describes the process as three steps: identify, invent and implement. He calls it a combination of creativity and discipline, not rocket science.

In the identify phase the team talks about behavior, which Placek says is bidirectional: how the brand behaves toward the market and how the market behaves toward it. He compares it to architecture, where people act differently toward a church than toward a Holiday Inn. Behavior leads into experience and into the rhythm of the name, as with the calming rhythm of Dasani.

The team also maps the competition's brand names and language, which Lexicon calls a landscape, because a name that isn't distinctive is imitation, which he calls a form of suicide. The output is a creative framework, which he deliberately avoids calling objectives because that sounds too logical. It works as a window for the linguists to travel through, so the results have breadth and different personalities.

Placek says the invent stage has two layers. The first runs against convention: Lexicon dropped large brainstorming sessions after finding that names came from small groups. On significant projects it uses three teams of two, each with a different briefing. One knows everything, one thinks it is working for a different company, and the third is naming something unrelated like a bicycle or car.

The point is to open up creativity: people working on a disguised assignment are free to make mistakes. Placek says most of Lexicon's names have come from the second or third team. Later he ties this to a general tip, noting that when something is intangible, like the code tool that became Windsurf, the team has to make it tangible.

The second layer is the part he says makes Lexicon unusual, the product of millions of dollars of R&D in linguistics and cognitive science. The firm has funded research on sound symbolism, worked with Stanford's linguistics department and employed 253 linguists over four decades. It now has a network of 108 linguists in 76 countries and databases of more than 18,000 small word units called morphemes.

After three to four weeks the team may hold two or three thousand ideas, not all workable. Paralegals and a trademark attorney screen them, because clearing a trademark is nearing a tipping point of difficulty. Linguists then narrow the list, and Lexicon usually shows clients two rounds, since humans like to compare, much as house hunters learn they want a view but not a pool.

Asked who the creatives are, Placek says the fundamental traits are curiosity, hard work and, ideally, low egos. He looks for tenacious people who keep producing when work is rejected, such as newspaper reporters, novelists and speechwriters. He found that agency people worked a little less effectively, and he says such people are not easy to find.

Naming a car is not a matter of everyone writing 100 names and hoping something falls out, he says, since not enough survives the legal and linguistic screens. Lexicon's software helps with sound symbolism, prefixes, suffixes and word units, so a team member can produce a list of 200 or 300 names in three or four days. He expects that in five years or so such software may produce solutions itself.

On timing, Placek says the ideal is eight weeks. Larger corporate projects with boards and more presentations take about three months, and some stretch to four once approvals and clearance are done.

Before it was Windsurf, the AI coding tool was called Codium. Placek says intangible things like a coding process are hard to name, so the team must make them tangible. One team was asked to list things, in real words like flow or in sports, that convey smooth, dynamic movement, and Windsurf was on that list. He says sometimes it is that simple, as long as the framework is right.

Placek calls Windsurf a compound, like Powerbook or Facebook, and says Lexicon's research shows compounds multiply associations: wind brings one set of images, surf another, so one plus one equals three. Clients often complain that compounds are long, so Lexicon did research on their effectiveness and showed the Codium team imagery they could use. He credits that team as intelligent and respectful, and says the name took about a week of back and forth to select.

Placek says AI naming differs in three ways. His clients are mostly engineers with little marketing background, the category moves faster than anything he has seen (the early internet looks like a daycare by comparison), and engineers want sophisticated names like Anduril or Anthropic. He spent about $20,000 on research with consumers and developers in Europe, South Korea and the US, finding developers positive and consumers skeptical and worried about their jobs. That is why he urged a more tangible, natural name than Codium, though he adds that Windsurf being available was partly luck.

Windsurf had just changed its company name from Codium, and Lenny notes that many AI firms use a different name for product and company. Placek says renaming gets harder as the customer base grows, so you must argue that the change is worth it. He names three situations where renaming makes sense.

First are startups that just needed a name for documents, perhaps while getting into Y Combinator and raising money, and that now realize the name is not very good. Second are companies that have pivoted, so the name no longer reflects who they are becoming. Third are mergers, where a new name signals that the company is new, bigger and more capable.

Lexicon's linguists mostly evaluate names rather than invent them, Placek says. They check for negative meanings, cultural or political associations and even local events, such as a flood in Italy, that a long-time US resident would not know about. He says that about every third or fourth project turns up something that is eliminated and never shown to the client, such as slang connotations from Australia.

On letters, Placek says V is the most alive and vibrant sound in English, which suits Corvette and Viagra. B is among the most reliable sounds, one rationale for Blackberry, whose founder thought Lexicon was crazy. Z is noisy, as in Azure, and X is fast and crisp, with semantic links to innovation.

Vercel was built from familiar fragments: *ver* as in veritas or verde, and *cel* as in accelerate, which matches what the company does for clients. Placek calls this processing fluency. Brains are a little lazy, he says, so complicated names are a liability and easy ones draw people in.

Lexicon is not a design firm, so its implementation phase is about helping the client team sell the name upward. That includes writing a longer rationale and building prototypes, such as the name on a baseball cap or T-shirt, or in a mock-up Wall Street Journal ad showing a rising share price. About half the time it also runs customer research, with drills meant to make people feel they are meeting a new brand.

Placek says the key research question is not whether people are comfortable with a name or whether it fits the concept, because that always produces a descriptive name. What matters is whether the name fires their imagination. He also tries to separate the client team from fear of their boss, since people want their boss to be happy, and encourages clients to lead the process and say they believe in the name.

Lenny asks about the 3,000-name figure. Placek clarifies these are ideas and directions, not ready-to-ship names, and the process is inefficient and a little chaotic. In his words there may be about 250 potential diamonds in that list that have to be examined.

For founders with only a couple of weeks, Placek says to forget about the name at first and draw a diamond. At the top write "win" and define what winning means. Then, clockwise, write what you have to win, what you need to win, and what you have to say to win. He says that if five people are in the room, they often have five different definitions of winning, and sorting that out matters.

The "have to win" corner covers what you already have that makes you a winner, which startups often underestimate. "Need to win" can include talent, resources or a distributor, and Placek says people often write "a good name," which he corrects to the right name. The last corner is what you need to say, including things you can't say yet, so the name stays flexible.

The exercise leads into behavior and experience, and he coaches people to look at metaphors, such as Blackberry telling the market it is not like the other guys, or Google versus Infoseek. It is not a one-hour task and may be repeated over four or five days. Then comes the generation stage.

If a team has a list of 200 names and feels nothing is there, Placek says 200 is not enough: aim for 1,000 or 1,500 names and directions, without evaluating them. He calls the industry overevaluating, since humans survive by spotting what is wrong with a picture, and he advises suspending judgment and asking what each name could do. His advice is ten or fifteen lists of 200.

Humans only notice what is different, Placek says, as with a red pair of shoes among black ones, so founders should look at what differs across their list and from the market. For help, he suggests asking employees what a name could do for the company, not whether they like it. Another drill is to tell friends that a new competitor has launched with the name, and listen to what it makes them imagine.

In consumer research, Lexicon knew it had a good name when someone said they knew little about the product but that it was not like the other guys. Placek says this creates a predisposition to consider it. He recalls that Andy Grove taught him that polarization signals strength: when the Intel committee was split between the descriptive ProChip and Pentium, Grove said that showed energy for Pentium.

On domains, Placek says the .com now matters about as little as a telephone area code, and AI will make SEO less important. Get the right name first, then add a prefix, a small word or a .ai address if needed, or buy the domain for perhaps $15,000 to $30,000 if you can negotiate. He would put the $30,000 into the market instead.

His closing tip is to think about behavior and experience, then use synchronicity: connect two unrelated ideas. A sailboat maker, for example, should skim hunting or flying magazines and note words they like, and Placek bets that one will relate to sailing. Lenny points out that this echoes the way Lexicon's disguised teams produce winning names.