A company in a new category has to teach the market two things at once: what the category is, and why this particular company deserves to belong inside it.
A founder launches a company in a market that does not quite have a name. The product is part software, part service, part research process. It helps marketing teams understand how AI answer systems describe their brand, but it is not exactly SEO software. It includes human review, but it is not a market research agency. It produces visibility reports, but it is not analytics in the usual dashboard sense. It touches reputation, but it is not classic reputation management.
The founder knows what it is. Early customers understand after a conversation. The sales deck has a diagram that makes the shape clear enough. The problem starts when the company has to exist without the founder in the room.
A directory asks for one category. The closest option is “SEO tools.” A journalist calls it “AI marketing software.” A buyer asks ChatGPT for “GEO agencies” and sees a list of SEO firms. Another asks for “brand perception research” and gets market research vendors. A third asks for “tools to track ChatGPT mentions” and expects a SaaS dashboard. The company appears in some answers and disappears in others, not because the offer is weak, but because the market does not yet know what shelf to place it on.
This is the problem of an unnamed category. Before a company can be evaluated, it has to be classified. In emerging markets, classification is often the hardest part.
Familiar categories are a quiet advantage
Companies in established categories get a benefit that is easy to underestimate. The market already has nouns for them.
A CRM company can say CRM. A payroll provider can say payroll. A help desk product can say help desk. The buyer may still need to compare features, pricing, integrations, and trust signals, but the first classification step is not especially difficult. Search engines, review platforms, software directories, analysts, procurement teams, and AI systems already have a shared vocabulary around the category.
A company in a new or blended category has no such luxury. It has to borrow language from adjacent markets while trying not to be swallowed by them.
This borrowing is risky. If the company borrows “SEO,” it inherits SEO expectations: rankings, keywords, technical audits, links, traffic. If it borrows “research,” it inherits expectations around interviews, surveys, reports, and qualitative methodology. If it borrows “analytics,” buyers expect dashboards, metrics, integrations, and ongoing measurement. If it borrows “reputation,” they think about reviews, crisis response, and sentiment.
None of these frames may be completely wrong. That is what makes the problem difficult. The company may touch all of them without belonging fully to any one.
AI search tends to expose this tension because it has to choose language on the user’s behalf. A system answering “who helps with AI visibility?” cannot preserve every nuance. It has to produce an answer that fits a category the user can understand. If the company has not made its category legible, the system may place it inside the nearest familiar bucket.
The first category wins more often than it should
A company’s first public category can become sticky.
This happens even in human markets. A startup launches as an agency because services are easier to sell early. Two years later it becomes software, but old customers, old articles, and old profiles still call it an agency. Or a company launches as a tool, then discovers the real value is a managed workflow, but public listings keep describing the tool. The first category becomes the ghost category.
AI systems can inherit that ghost. If old pages, directories, biographies, and comparison articles all use the original label, newer language has to work harder to displace it. The company may be commercially different now, but the public archive still has weight.
The same thing happens when the company uses a temporary category as a bridge. It says “SEO” because buyers know SEO, then later tries to move toward “AI visibility.” It says “research” because the early deliverable is a report, then later tries to explain operations. It says “platform” because it sounds more scalable, then has to clarify that customers do not log into a self-serve product.
These early compromises are often rational. The risk is that they become public facts.
This is one reason emerging-category companies need to be unusually careful with public profiles. A directory field filled in casually during a rushed launch can become a durable source. A founder bio written for a podcast can become the clearest description of the company available online. A launch article can outlive the launch strategy. The web has a long memory for provisional language.
Differentiation cannot come before placement
Founders often resist plain category language because it feels reductive. They do not want to sound like another SEO tool, another analytics vendor, another research agency, another AI consultancy. They want the first sentence to express what is new. The impulse is understandable. It is also dangerous.
A buyer cannot appreciate differentiation before they understand placement. The same is true for AI systems. If the company begins with a phrase like “decision infrastructure for the AI-mediated enterprise,” the reader may sense ambition but not category. If the page never descends into ordinary nouns, the company becomes hard to compare. Placement is not commoditization. It is orientation.
A good first sentence may feel almost disappointingly plain. It says what the company does, who it serves, and what problem it addresses. The more distinctive thesis can follow. In fact, it lands better after the reader knows where they are.
A company might say: “We help B2B teams monitor and improve how AI search systems describe their brand.” That sentence will not win a poetry prize. It does useful work. It gives the buyer a category-like frame, even if the category is still forming. Once that frame exists, the company can explain that the work combines AI visibility audits, public-source cleanup, perception research, and trust evidence. Nuance has somewhere to attach. When differentiation comes first, nuance floats.
New categories need repeated boring facts
The more novel the company, the more repetitive its factual layer has to be.
There is no need to repeat marketing copy word for word across every page. The basic public facts, however, should not drift: the company type, audience, problem, service model, output, and boundaries. A buyer should not see “AI visibility infrastructure” on the homepage, “SEO analytics” on LinkedIn, “brand research” in a directory, and “reputation management” in an article unless the company has deliberately explained how those frames relate.
Repeated boring facts are not a branding failure. They are category scaffolding.
The public web needs enough stable language to build a coherent version of the company. AI systems need it too. Google says AI Overviews and AI Mode may use query fan-out across subtopics and sources to develop responses, which means multiple public surfaces can contribute to how a topic or company is assembled. Google Search Central describes this multi-search behavior as part of the AI search experience.
If those surfaces use incompatible category language, the system has to resolve the conflict. It may resolve it badly. Buyers may do the same.
This is why category work is not only a positioning exercise. It is source-trail work. The company has to make the emerging category visible across the places where buyers and systems learn: service pages, about pages, case studies, directories, review platforms, partner descriptions, founder bios, public interviews, and comparison content.
The category is not real enough if it exists only in the sales deck.
Adjacent categories are useful, but they should be handled openly
No emerging category grows in isolation. It needs neighbors.
A company working in AI visibility may need to explain how it differs from SEO, GEO, digital PR, market research, reputation management, and analytics. A company in agentic commerce may need to explain how it differs from e-commerce search, product feed optimization, shopping ads, and marketplace operations. A company in human-signal research may need to explain how it differs from user testing, traffic generation, and fake engagement. The mistake is pretending adjacency does not exist.
Buyers will compare you with familiar categories whether you like it or not. AI systems will do the same. If the company does not explain the relationship, the comparison will be made by a directory, a competitor, a model summary, or a buyer’s first impression.
A mature category page does not only say “we are different.” It shows the boundary. It says where the adjacent category is useful, where it stops, and why this newer category has become necessary. The tone matters. If the company dismisses every neighbor, it sounds defensive. If it refuses to draw boundaries, it sounds vague.
The best explanations are generous but firm. SEO still matters, but AI visibility asks a different question. Market research still matters, but public-source interpretation is a different surface. Reputation management still matters, but source-trail coherence is not the same as crisis response.
This kind of writing helps buyers. It also gives AI systems better material for comparison answers.
The category may be forming in prompts before it appears in reports
Traditional category formation often showed up in analyst reports, software directories, venture decks, conferences, and procurement language. Those still matter. But AI search adds a strange new early signal: prompts.
People ask for a thing before the thing has a stable name.
They ask why ChatGPT describes their company incorrectly. They ask how to appear in AI answers. They ask why competitors show up in Perplexity. They ask how to track brand mentions in AI search. They ask whether AI visibility is different from SEO. They ask how to clean up sources that AI uses.
Each prompt is clumsy, but together they outline an emerging need.
This is where companies in new categories should pay attention. The market may not yet use your category name, but it may already be asking your category’s questions. Those questions can become content, service pages, product language, and public definitions. They can also reveal when the proposed category name is too abstract for the problem buyers actually feel.
A category name is useful only if it helps people ask better questions.
The name may arrive late
There is a temptation to believe that naming the category solves the category problem. It rarely does.
A name can help, but only when the market has enough experience to attach meaning to it. Before that, the name is a handle without a suitcase. It can be repeated, but it does not carry much.
This is why early category work should not become obsessed with the label. The label matters less than the public explanation beneath it. What problem is becoming more common? What old category fails to solve it? What new behavior is emerging? What evidence shows that buyers already care? What does a company in this category actually do? What should it not promise?
The category name may change. The underlying explanation should get clearer.
AI search will reward that clarity before it rewards the perfect label. A system does not need the category to be famous in order to describe it well. It needs enough public material to understand the relationship between the problem, the company, the neighboring categories, and the evidence.
For a company in an unnamed category, the first job is not to sound revolutionary. It is to become classifiable without becoming flattened.
That is a harder writing problem than most founders expect.