Chapter 1 said: if you can't be first, create a category you can be first in. This is the engineering manual — how new categories actually get built, and the gap in the adoption curve where most of them die.
Here's the whole chapter in one line: a category is born when a problem gets a name and a budget line — and it survives only if someone carries it across the gap between the people who love new things and the people who buy safe things. Everything below is the mechanics of that sentence.
Chapter 1 gave you the ladder: for every category, the mind keeps a short ranked list, and the cheapest move in marketing is to claim an empty one. Chapter 3 filed the law of the category under "survived the evidence." What neither chapter said is what a category actually is — and the answer is more bureaucratic than the books make it sound.
A category exists when two things exist at once. First, a mental slot: buyers can say the words and picture the shape of the thing — "CRM," "password manager," "espresso machine." Second, and this is the part that pays salaries, a budget line: somewhere in a spreadsheet, money is pre-allocated to the words. The ladder in the head, institutionalized into a row in the plan.
Programmer's version: a category is a namespace the budget process recognizes. Until finance can bind a purchase order to the name, your product is an unbound symbol — every deal starts with a linker error. Positioning inside an existing namespace is cheap: the resolution machinery already exists, you're just arguing about which value the name points to. Creating a namespace means shipping the resolver too.
Which is why the economics of this chapter cut both ways. Nobody buys the fourth-best anything — you learned that on the ladder. But plenty of people buy the first something-new, provided someone has done the slow, expensive work of making the something-new feel like a thing that sane people budget for. That work is category design, and the rest of the chapter is its bill of materials.
Every pitch deck has the bell curve: innovators (2.5%), early adopters (13.5%), early majority (34%), late majority (34%), laggards (16%). Most decks read it as a demographic parade — cool people first, boring people later. That reading is wrong in the way that costs money. Chapter 4's rule applies here: segment by the buying logic, not by the person. The five segments are five different questions, asked before money moves.
Innovators ask "is it new?" They'll install anything once, file bug reports for fun, and pay mostly in feedback. Early adopters — Moore's visionaries — ask "is it a breakthrough?" They buy change itself: a chance to leapfrog their industry before it's proven. Rough edges are almost a credential; if it were polished, it would already be everybody's.
The early majority — the pragmatists — ask a different species of question: "who else like me uses it?" They don't buy change; they buy references and whole products. Then the late majority buys when not buying has become the risky choice, and the laggards buy when the old thing is discontinued.
Programmer's version: the curve runs from users who tolerate beta to buyers who demand LTS. A visionary will run your nightly build in production and brag about it. A pragmatist wants a long-term-support release, a migration path, three customers on the same stack who'll take their call, and someone to blame. Same product; incompatible acceptance criteria. The mistake is assuming the segments differ in speed — that pragmatists are just visionaries on a delay. They're not slower. They're running a different test suite.
In 1991, Geoffrey Moore looked at a decade of tech startups that grew fast, raised big, and died anyway, and found they all died in the same place: between the early adopters and the early majority. He drew the bell curve with a crack in it and called the crack the chasm. The book — Crossing the Chasm — became tech marketing's physics text because it explained a thousand post-mortems with one picture.
Here's the mechanism, stripped to the deadlock. Pragmatists buy references — but only references from other pragmatists. A glowing quote from a visionary is worse than useless; to a pragmatist, visionaries are the people who buy things that don't work yet. So the first pragmatist wants proof from a customer who cannot exist until some pragmatist goes first. It's a bootstrapping deadlock: a circular dependency at link time. Module A imports proof from module B, module B imports proof from module A, and nothing loads.
The cruelty is in the timing. The chasm arrives disguised as success. You've sold every visionary in reach; revenue is up and to the right; the board approves a bigger number. Then the visionary pool — remember, 13.5% at best — runs dry, and the pragmatist pipeline that was supposed to replace it produces enthusiastic meetings and no purchase orders. Momentum dies with the lights on.
Moore's escape from the deadlock is a war metaphor he meant literally: D-Day. The Allies did not invade Europe evenly. They took one beach, held it at terrible cost, and used it to land everything else. The chasm version: stop selling to "the market." Pick one narrow niche — narrower than feels safe — and win it completely.
Completeness is the whole trick, and it has two parts. Part one is the whole product: not your core product, but everything a pragmatist needs to get the promised result — the integrations, the onboarding, the training, the support contract, the migration path. Programmer's version: pragmatists don't install your library, they install its transitive dependency closure. Visionaries will vendor the missing pieces themselves; pragmatists take the closure or nothing. Inside one narrow niche, finishing the closure is achievable. Across "the market," it never is.
Part two is reference density. Pragmatists in a niche all know each other — same conferences, same group chats, same three consultants. Win six of the forty companies that matter and the other thirty-four hear about it without you in the room. The deadlock breaks because the reference a pragmatist demands finally exists and is someone they golf with. Word of mouth has a critical mass, and it's only reachable in small ponds.
Then — only then — you bowl to the adjacent pin: the neighboring niche that shares buyers, references, or whole-product pieces with the one you own. Each conquered niche is a pin that knocks into the next. This is Chapter 7's principle of force in adoption-curve clothing: you cannot be superior everywhere, so be locally superior — pick a battlefield small enough that your whole company outweighs the incumbent's rounding error.
The killer is the temptation to stay broad. Narrow feels like giving up revenue; the TAM slide is right there. But broad across the chasm means being a stranger everywhere — every deal a first-sale, every vertical a cold start. The niche isn't the prize. The niche is the bridge.
Moore tells you how to cross into a category. Play Bigger (2016, Ramadan, Peterson, Lochhead & Maney) tells you how to build one worth crossing into. Strip the airport-book gloss and the playbook is four moves, in order.
Move one: name the problem, not your product. Buyers don't budget for solutions to problems they haven't named. So before anything else, category designers give the pain a name and an enemy — something buyers can grumble about in meetings you're not in. Move two: publish the POV — a point of view that says the old way is broken, here's why, here's what the new way is called. A manifesto someone can forward. Move three: design the ecosystem — partners, integrations, consultants, courses; a category with one vendor is a product with a press release. Move four: lightning strikes — concentrated bursts of attention (a conference, a report, a launch) that make the category feel inevitable rather than argued.
The clean case is HubSpot. Two founders with a marketing tool in 2006 did not say "buy our marketing tool." They named the enemy — interruption marketing: the cold calls and spam everyone already hated without a word for it — and gave the alternative a name: inbound. Then they taught it, free, for years: blog, certification, an academy, a conference called Inbound before the product deserved one. By the time competitors noticed, "inbound marketing" was a budget line, and the company that named it was, per Chapter 1's oldest law, first on the ladder it had built.
Gong ran the quiet rerun a decade later: hammer the problem ("deals die in calls nobody reviews"), then christen the category — revenue intelligence — and feed it data-driven content until analysts wrote the name down. The order is the lesson. Problem first, category name second, product last. Run it backwards and you're just another vendor shouting a noun.
Why go to all this trouble? Because categories don't split their winnings evenly. The Play Bigger authors studied U.S. tech startups and claimed that category kings — the company that defines and leads a category — capture roughly two-thirds to three-quarters of the category's total market value. Not the best product: the company the category is about. Salesforce is what CRM means; the rest of the market shares the leftovers. Winner-take-most is the gravity of namespaces: once a name resolves to you by default, every mention of the category — by anyone, including competitors — is advertising for you.
Between the namer and the crown sits the analyst-industrial complex. A category isn't fully real in enterprise software until Gartner draws it a Magic Quadrant or Forrester writes it a Wave — a quadrant is a budget line's birth certificate. Once it exists, procurement can justify the spend, CIOs can cite it, and the deadlock of §3 loosens: the analyst is a reference pragmatists accept by proxy. Which is why category designers court analysts the way startups court investors, and why a freshly named category with no analyst coverage is still, financially speaking, a rumor.
The third mass in the system is community. HubSpot Academy certifications, Salesforce's army of admins with the category on their résumés, user conferences that feel like family reunions — these look like marketing programs and function as moats. When thousands of people's careers depend on your category existing, the category defends itself. Gravity, once assembled — default name, analyst certificate, career ecosystem — is the closest thing marketing has to a durable monopoly. Which is exactly why the claims around it deserve an audit.
Category-design literature is written by the lottery winners. For every HubSpot there is a graveyard of well-funded category launches, and the graveyard sorts into three failure modes worth memorizing.
Mode one: the education bill exceeded the runway. Teaching a market to grumble in your words takes years of content, conferences, and patience — HubSpot ran the school for the better part of a decade. Burn rate doesn't care how good the POV is. Mode two: too early. "Push technology" was 1997's inevitable category; PointCast got the magazine covers and died waiting for the world to need it. A category can be correctly designed for buyers who don't exist yet. Mode three: an incumbent absorbs the word. You can spend years making a term famous and watch a giant staple it onto their bundle — ask anyone who built an "AI copilot" before Microsoft made Copilot a product line across an entire empire. The namespace was claimed; the resolver was theirs.
The audit yields a practical rule: sometimes the right move is to rename an existing category, not create one. If a budget line adjacent to your product already exists, repositioning within it — "next-generation X," "X without the servers" — borrows a working resolver instead of shipping a new one. Category creation is only worth the bill when buyers genuinely grumble about a problem no line item pays to fix.
The 2023–26 AI land-grab is the live experiment. Dozens of startups have declared themselves kings of "agent ops," "LLM observability," "AI SDRs," "autonomous finance" — a fresh category every funding announcement. The budget lines underneath are still consolidating; CFOs are collapsing seventeen claimed categories into two or three rows called something like "AI tooling." Most of the names will not survive the consolidation. The ones that do will belong to companies that ran this chapter in order: beachhead first, problem named, references dense — then the crown.
Fold Moore and Play Bigger together and you get one checklist. It's sequential — each item is load-bearing for the next, and the classic failure is running item seven while item three is still false.
One thread runs through every item: words. The problem gets a name, the category gets a name, the enemy gets a name — and then your product, and everything else you ever ship, needs names that don't squander what the first three built. That's the next chapter's territory, and it's where more equity gets burned than anywhere else in marketing.