Check your positioning against real searches

Naming usually finishes before research begins. The market position is picked, the tagline is drafted, the homepage repeats it, and the keyword check happens afterwards if it happens at all. By then the language has been argued for, designed around, and paid for, which is precisely why nobody enjoys hearing what a search box has to say about it.

The check takes minutes. Run it while the words are still cheap to change and you get a different answer than running it once they are everywhere.

What the test can return

Look up the phrase you use for what you sell, along with the obvious variants, and read two things: whether anybody searches it, and what the pages that do rank are offering.

Three results are possible.

Demand exists and the ranking pages sell what you sell. You are working with language people already use, and the job from here is building pages against it.

Demand exists and the ranking pages answer something else. This is the expensive one. The phrase is in use, but it means a different thing to the people typing it, and their meaning is older and more established than yours. A page built on that phrase competes against a definition it cannot displace.

Nothing. The most common result, and the one least planned for.

What an empty result is worth

A phrase nobody types still carries information, because the searches sitting near it usually describe the same need in the words the market settled on. Those are often one query away, and reading them is the fastest way to acquire the vocabulary you skipped. The failed naming exercise hands you the replacement for free if you look one step past it.

The invented term keeps its job. It is ownable, it is memorable, and it gives a company a way to talk about itself with a straight face. The error is asking it to be findable as well. Coined language belongs where you get to explain it — the product, the pitch, the onboarding, the sales call. Established language belongs on the pages someone lands on without any help from you.

Low volume is a separate finding from none. A term with a little volume, strong intent, and no serious competition can be worth holding, and treating those two results as the same thing discards good terms while keeping bad ones.

When a phrase is being pushed from outside

Occasionally a phrase acquires searches because somebody is funding them: a national campaign, a trend piece, a category another company is buying its way into. If what you sell genuinely matches, arriving early is not derivative. The demand was created on someone else's budget and the searcher still needs somewhere to land.

Seeing this early means watching the language entering your market from outside it, not only listening to what customers say in calls. Seeding from conversation and this are the same discipline facing opposite directions, and running both is how you find vocabulary you neither invented nor heard directly.

Four steps to run before the site is written

1. Check the term and its variants for volume. An empty result is information, not an inconvenience.
2. Search it and read the first page of results. If what ranks there solves a problem you are not solving, the phrase belongs to an intent you will not take.
3. List the phrases from your last ten sales conversations and support tickets. That is the demand side, stated by the people doing the demanding.
4. Put the two lists side by side. Where they disagree, the buyers' version carries the pages a stranger arrives on and yours stays where context already exists.

The fourth step is where the argument happens. The buyers' phrasing is plainer, and plainer sounds like less. It is also what a stranger can type from a standing start, which is the only test that matters for a page that has to be found.

Do it with an agent

Nothing here needs a manual pass. Connect the Doverity MCP and this prompt does the check, the comparison and the recommendation in one run.

Use the Doverity MCP. Call get_keyword_metrics on "[my term]" and on
these variants: [list], and give me the volume, difficulty and intent
numbers for every one.

A. Where a term comes back thin or empty, run research_keywords
around it and name the nearby terms that do get searched. Say what
the pages ranking for those terms are offering.

B. Here is the wording buyers used for the problem, taken straight
from calls and tickets: [paste]. Match each phrase to a keyword where
one exists, and flag the ones with no keyword behind them.

C. List where my wording and theirs part company, and mark which of
theirs belong on public pages. Mine stays in the pitch and the
product, where I get to explain it.

The last line matters. The point is not to abandon the language you chose. It is to stop expecting that language to be found.

Positioning and search demand FAQ

Is it worth coining a term for what you do?

It can be, if you want to own the word and you accept that it starts with no searches behind it and may never gain any. A coined term needs a plan that does not depend on being discovered, which generally means direct distribution and a lot of explaining. Anything a newcomer has to look for still needs to be phrased in words that exist already, so budget for two vocabularies rather than one.

Can a keyword with no volume still be worth targeting?

Only after you have worked out whether it is truly empty or merely small, because those lead in opposite directions. Genuinely empty on a category phrase usually means the market names the problem differently, so the next move is finding the neighbouring terms that do get searched and reading what ranks for them. Small volume with strong intent and no competition is often worth holding. Volume was never the only reason to want a phrase.

Which records show how buyers describe their problem?

Sales calls, support tickets, and interview recordings. Nobody describes their problem in the vocabulary of a keyword database, which only holds phrases people have entered often enough to show up at all. Read the original wording, before anyone in the company paraphrased it into a feature request.

Is a missing search volume number a bad sign?

It is a sign about language, not about demand. Things people pay for exist outside search entirely, because nobody thought to look them up that way. What an empty number does tell you is that search cannot be the channel carrying it, and that the words which would carry it are worth finding before the site gets written.