What category should our B2B product be in?

Your B2B product should be in the category your buyer already types into ChatGPT, Google and the procurement form. Then you name the problem you solve inside that category, in the buyer's words, so you get filed where the buyer is looking and stand out once they get there.

A category label your buyer has never searched for gets you filed nowhere. Humans skip it. AI engines, which now answer a growing share of "who should we look at for this" questions, have no shelf to put you on.

Why does the category question feel so hard?

Because two true things pull against each other. Your product is different, and difference feels like it deserves its own name. But buyers shop with the words they already have. A hospital CFO doesn't type "revenue intelligence orchestration layer." She types "claims denial software" or "how do we stop losing cash to denied claims."

The buyer's words are the shelf. Your difference is the label on the box.

April Dunford gets this right in Obviously Awesome. Market category is one of her five positioning components, and she chooses it last, after competitive alternatives, unique attributes, value and best-fit customers, so the category frames your strengths instead of burying them. Her three styles, head-to-head, big fish in a small pond, and create a new game, are the honest menu. What I'd add for a $5M-$75M company in 2026 is a check she didn't need to write in 2019: where do the AI engines already file you?

How do I choose the right category for my SaaS product?

Run four tests, in this order, before you commit a word to the homepage:

1. 1The search test. Pull your last 20 inbound leads and your win/loss notes. Write down the exact nouns buyers used for the thing they were shopping for before they found you. Those nouns are your candidate categories. If none of your candidates appear in that list, you're holding a name buyers haven't learned yet.
2. 2The comparison test. Ask five recent buyers who else they looked at. The companies they name tell you the shelf you're already on, whether you like it or not. Pick the category that makes that comparison set work in your favor. This is Dunford's point exactly, and it still holds.
3. 3The filing test. Ask ChatGPT, Claude and Gemini two questions: "What are the best [candidate category] tools for [your buyer]?" and "What does [your company] do?" If you're absent from the first answer and misdescribed in the second, the engines have filed you somewhere else, or nowhere. That's the category you're really in today, whatever the deck says.
4. 4The problem test. Name the problem you end, inside that category, in seven words or fewer, in the buyer's words. "Claims denial software" is a shelf. "Claims denial software for hospitals that can't hire more billers" is a shelf plus a problem. The problem is what makes you findable inside the category instead of interchangeable with the rest of it.

Run all four and the answer is usually the boring one: an existing category your buyer already uses, with a sharper problem statement than anyone else on that shelf.

What market should we compete in?

The one where three things line up. Buyers already have budget and words for the problem. You have proof you end it. And the comparison set on that shelf is weakest on the exact problem you end. "Where can we win" is a question about the comparison set, and the search test plus the comparison test answer it faster than a strategy offsite.

Here's a hypothetical to show the reasoning. A scheduling product built for home-health agencies could compete in "workforce management" or in "home health software." Workforce management is the bigger market, with bigger budgets, and the comparison set is a dozen generalists who have never met a visiting nurse. Home health software is the smaller shelf, buyers search it by name, and the comparison set is EHR vendors who treat scheduling as an afterthought.

The second shelf wins. The buyer walks to it on their own, and the product is the only box there that names the scheduling problem. That's big fish, small pond, with the problem test deciding which pond.

What does our own data say about inventing your own category language?

We scored 302 healthtech homepages on 19 criteria for the 2026 Healthtech Messaging Index. 63 of them earned full marks for owned language and category: they coined a term, trademarked a product name or framed a category of their own. Only 29 of those 63 also passed the 7-second test, where a stranger can tell who the page is for, what problem it solves and what it stands for within seven seconds. The other 34 coined the language and lost the stranger.

The 29 that did both averaged 26.8 of 38 points. The 34 that coined and confused averaged 22.3. That's a correlation inside one vertical, and it matches what I hear on calls: the invented category tends to arrive before the plain-words problem, and the plain-words problem is the part a buyer and an AI engine can actually file.

Coining a term costs you nothing if the problem is already named in the buyer's words. Coining it instead of naming the problem costs you the stranger.

Should we create a new category instead?

Sometimes, and rarely at $5M-$75M. Creating a category means teaching the market a new word before anyone will search for it, and that teaching bill runs for years. I walked through the difference in Category design vs positioning: which does your B2B company need?, and if you want to try it without a Play Bigger budget, the six-step version is in How do you create a new market category?.

Create a new category only when the search test comes back empty because buyers truly have no word for the problem, and you can fund the teaching. Otherwise, name the problem inside the existing category and own that sentence. The company that names the problem gets presumed expertise inside the category, which is the whole argument of why the company that names the problem owns the market.

Most companies that think they need a new category need a sharper problem statement inside an old one.

How do you make the category stick once you've picked it?

Write it down once. The category, the comparison set and the seven-word problem sentence go in one place every rep, every page and every AI tool reads from. At PitchKitchen that place is the Magnetic Messaging Framework, and the AI Brand Twin trained on it is why a client's ChatGPT describes them the same way their best rep does. Left undocumented, the category drifts back toward whatever the last competitor's website said, which is the same failure I covered in How do we position a product in a crowded market?.

Then re-run the filing test every quarter. Engines re-read the web constantly, and a category you held in March can be re-shelved by September if three competitors start using your words.

Want a fast read on whether your category and problem land today? The free Brand Signal Score scores your homepage the way a stranger and an AI engine read it, in a few minutes, and the 7-second test is the first thing it checks.

Pick the shelf the buyer walks to. Then be the only box on it that names their problem.