Every market runs on attention. But attention is not the same thing as understanding.
The distinction matters because a message can attract attention without creating understanding. Customers can hear a message and miss the point. Investors can hear a strategy and draw the wrong conclusion. Analysts can understand what a company does and still misunderstand why it matters.
Markets don’t read minds. Rather, they respond to what they can understand. That observation turns out to have an unexpected connection to the origins of modern AI.
Attention please
In 2017, a team of Google researchers published “Attention is all you need,” the seminal paper that introduced the Transformer architecture. The title has since become part of the mythology of artificial intelligence, invoked in nearly every explanation of how large language models came to be.
But the original context is easy to miss. The paper was not written as a grand theory of intelligence. It was a contribution to machine translation.
That detail matters. In ordinary language, attention means focus: what people notice, prioritize, and spend mental energy on. Phrases like “the attention economy” capture the modern reality that attention is a scarce human resource. Media platforms, brands, politicians, and creators compete to capture it.
In the Transformer paper, attention meant something more technical. It was a way for a model to weigh relationships among words. To understand one token, the model learned which other tokens mattered. Meaning did not emerge from each word in isolation, or from a sentence moving forward in a straight line. It emerged from context.
The two meanings are not the same. Machine attention is not human attention. But they address the same challenge: not everything matters equally. Successful translation depends on knowing what deserves focus, what can be ignored, and how the parts fit together.
In machine translation, the challenge is carrying meaning across languages. The same challenge appears in markets.
Beyond language
Translation is not just a language task. It's how ideas cross borders between markets, cultures, institutions, and disciplines.
Translation is how a technical breakthrough becomes a product. It's how a product becomes a category. How a category becomes a budget line. How a company’s internal conviction becomes something customers, investors, analysts, employees, regulators, and journalists can understand and use.
Most organizations underinvest in translation because they misunderstand what it is. They treat it as simplification, packaging, messaging, communications, or localization. Those things matter, but they are all downstream applications.
Every serious company is in the business of moving ideas across borders. Some of those borders are geographic. Most are not.
Real translation happens earlier and goes deeper. It asks: What does this idea mean in the world of the person who needs to understand it? What must be preserved? What must be adapted? What assumptions will not travel? What will be misunderstood if we leave it in our own language?
A company doesn’t define a market by saying the same thing as everyone else, only louder. It defines a market by making a new idea legible to the people whose recognition gives the market shape.
Captive ideas
Inside companies, ideas often begin in expert communities, each with its own language.
Engineers speak in systems and architecture. Product leaders speak in capabilities. Founders speak in conviction. Researchers speak in evidence. Finance speaks in models. Sales speaks in buyer resistance. Legal speaks in risk.
None of these languages are wrong. But none of them automatically travels beyond the community that created it.
That's where many ideas get trapped.
A product described internally as architecture may need to be framed as workflow for a buyer, market expansion for an investor, category definition for an analyst, strategic position for a board, and news for a journalist.
Each audience asks a different version of the same question:
Why does this matter?
If the idea can't answer that question in the language of the audience, it remains captive inside the company.
That’s a translation problem.
Lab to market
The researchers at Xerox PARC helped invent many of the foundational technologies of personal computing: the graphical user interface (GUI), the mouse, and Ethernet. But Xerox never fully translated those inventions into a market the company could own.
In 1979, Steve Jobs led a team of Apple engineers and executives on a now-legendaryvisit to Xerox PARC. They saw the GUI and understood that it enabled a new relationship between humans and computers.
Naming a category is easy. Making it mean something is hard.
Apple then did something Xerox had not managed to do at scale: It translated innovation into use, desire, product, story, and distribution.
Takeaway: Markets don’t reward the pure idea. They reward the idea made legible.
The meaning of a milkshake
Clayton Christensen’s “Jobs to be Done” framework makes a related point from another angle.
His famous milkshake example is funny because it sounds absurdly small. A fast-food chain wants to sell more milkshakes. They study demographics and flavor preferences. But the real question turns out to be different: what job is the customer hiring the milkshake to do?
For some morning commuters, the milkshake is breakfast, entertainment, convenience, and a way to make a dull drive feel tolerable. Once the company understands that context, the product looks different. So does the competition.
The milkshake is not only competing with other milkshakes. It's competing with bananas, bagels, coffee, boredom, hunger, and the awkward fact of driving to work with one hand free.
That is translation. It’s what moves a product from the company’s language into the customer’s life.
Markets are social
The same pattern shows up in the diffusion of innovation.
New ideas don’t spread through markets as pure information. They move through social systems. People adopt what they can understand, trust, justify, and see others using.
Early adopters don’t think like the early majority. Technical buyers don’t think like economic buyers. Enthusiasts don’t think like skeptics. Every innovation must be translated anew as it crosses each of these social boundaries.
This is one reason category creation is so difficult.
Naming a category is easy. Making it mean something is hard.
A company may have a useful, innovative product and still fail to create a convincing frame for it. Customers hear one story. Analysts hear another story. The sales team improvises a third. The website says something broader. The board deck says something more cautious. The product roadmap implies something else altogether.
In this very common scenario, the company doesn’t have a category. It has fragments of different stories that don't quite fit together.
Have words, will travel
Category strategy is the discipline of moving ideas across audiences.
It’s not wordsmithing, or the search for a slogan. It’s the work of making an idea intelligible and useful to different constituencies without reducing it to mush.
The best category language has this quality: it travels. It can be used by a founder in a board meeting, by a salesperson in a customer conversation, by an analyst in a market note, by a customer in an internal business case, and by an employee explaining why the company matters to family and friends.
Each use case is different and requires specific language. But the underlying idea remains the same.
That is the test. Not whether the language sounds polished in the abstract. Whether it survives contact with the audiences that matter.
The fluency trap
Fluency is not the same as translation.
AI has made fluency abundant. Anyone can now produce a plausible paragraph, a confident email, a tidy summary, a marketecture diagram, a thought leadership draft, a synthetic customer persona.
Fluency was the old scarcity. In a world of unlimited fluency, translation is the new scarcity. Fluency moves words. Translation creates understanding.
AI can restate a concept in simpler terms. It can vary tone, format, length, and vocabulary. That's useful. But translation is not just changing the words. It requires a theory of the audience, knowledge of the world into which the idea is being carried, and an understanding of power, context, incentives, anxiety, aspiration, and consequence.
This is where the work becomes strategic.
Translation is not about dumbing down your ideas. It's about proving that you understand the people you're trying to reach. That means showing respect for their context by doing the work required to be understood.
Locked rooms
When translation fails, companies often retreat into jargon.
Jargon is not simply ugly language. Sometimes jargon is necessary shorthand inside expert communities. The problem comes when internal shorthand is mistaken for external meaning.
Jargon then becomes a locked room. It reassures insiders that they understand one another while leaving everyone else outside the door.
Here's the rub: Markets don’t reward what you meant. They reward what others can understand, believe, and use.
I learned that lesson long before I worked with technology companies. As an anthropologist in Lahore, I quickly realized that translation was never just a matter of words. It involved class, religion, kinship, humor, obligation, colonial history, gender, status, and all the invisible context that determines whether a sentence lands as intended.
You could know the dictionary meaning of every word and still miss the meaning of the exchange.
Business has its own versions of the same problem. A boardroom, a sales call, an analyst briefing, a policy debate, a product meeting, and a customer workshop are all cultural settings. Each has its own codes, silences, taboos, incentives, and forms of authority.
To speak clearly, you must first understand how meaning is made.
Border crossing
Every serious company is in the business of moving ideas across borders. Some of those borders are geographic. Most are not.
The more complex the market, the more important translation becomes.
AI makes this challenge more urgent. As fluency becomes commodified, the ability to create real understanding becomes more valuable. Companies won’t win by producing more clever, plausible content. Anyone can do that. The winners will be those that succeed in making difficult ideas travel successfully across markets and communities.
Translation matters because the world is defined by borders.
Most of them don’t appear on maps.