Keyword research that tells you what to build next

What keyword research is, and what sends an owner looking for it

Keyword research is a demand survey for one specific business. It is not a list of words. It answers a question an owner cannot answer from the inside: out of everything you sell, in the places you will actually drive to, what are people typing and asking, and which of those searches are worth a page. The output is a decision about where the next several months of work and money go.

Almost nobody calls and asks for this by name. They call because one of these is happening.

• You are about to pay for pages and nobody can say which pages. The proposal has a number of pages per month in it and no answer to which ones, or why those.
• The site ranks for your company name and little else. People who already know you find you. Nobody else does.
• Traffic went up and the phone did not. The pages you have pull in readers who were never going to hire anyone.
• A previous agency left a pile of near identical city pages. You cannot tell which are worth keeping, which need rewriting, and which are quietly hurting you.
• You changed what you sell. You added a service, dropped one, or stopped taking the small jobs, and the website still sells the old business.

Underneath all five is the same problem. The site was built on an assumption about demand that nobody ever checked.

What lands in your hands

This is a deliverable, not a phase you hear about in a status update. When it is done you have files. Here is what is in them, specific enough that you can check you got it.

• A keyword map. Every search worth your time, grouped into clusters, one cluster per page. Each row names the page that should own it, either a live URL or a note that it has to be built, a priority tier, and one line on why it made the list.
• A build order. The first pages we would build or rewrite, in sequence, so the work has a starting point on Monday instead of a spreadsheet nobody opens twice.
• A do not build list. The searches we left out on purpose, each with the reason: too broad, wrong buyer, wrong town, out of reach for the site as it stands, or a job you told us you do not want. Most owners never get handed this half, and it usually saves more than the other half makes.
• An overlap report. Where two or more of your existing pages chase the same search and split the difference, with a call on which one should win and what happens to the other.
• A question inventory. The full sentences people say, not only the two and three word phrases: what they ask on the phone, and what they describe to an assistant that answers in a paragraph.
• A short written brief. Plain English, so you can hand it to a writer, a new marketing hire, or the next person who tries to sell you SEO, and watch how they react to it.

It arrives as a spreadsheet and a document in your own account, yours to keep. Not a login to a dashboard we switch off when the engagement ends.

How we decide what makes the list

Turning a raw pile of phrases into a plan takes judgment, and most of that judgment is about your business rather than about search. Four things decide it.

• What the job is worth, and whether you want more of it. As a worked example, say you run a plumbing company. Drain clearing gets searched constantly and bills a few hundred dollars. Sewer line replacement gets searched far less and bills thousands. Which one deserves the first page is a question about your margins and your crew, not about which phrase has the bigger number beside it. That is why the owner sits in the room for this part.
• How close the searcher is to hiring. Some searches come from a person standing in water. Some come from a person reading on a lunch break who might buy in March. Both can deserve pages. Not the same page, and not in the same order.
• Whether the site you have can realistically win it. A search held by four national brands is a different project from one held by a competitor with a slow site and a thin paragraph. We look at what stands in front of each search before putting a page against it, and the deeper version of that question is its own work: competitor analysis.
• What you can actually handle. If you can take four more jobs a month, a plan built for forty is a fantasy that costs real money. Better a short list you finish than a long one that makes the invoice look impressive.

One filter runs underneath all four. Some searches are not won with a page at all: the ones settled on the map, where proximity and your Google Business Profile decide more than wording ever will. Those get flagged and handed to local SEO instead of being assigned a page.

How the work actually runs

It starts with a call, not a form. We have to hear what you sell before we can find out who is asking for it.

What we need from you. Your real service list, including the ones that never made it onto the website, and a rough sense of which are worth the most relative to the others. The areas you will drive to on a Tuesday, and the ones you take only when the job is big enough. The jobs you do not want, which shortens the list faster than anything else. What people ask when they call, in their words, because your phone is the best source of question data a local business has. Read access to Search Console, analytics, and your Business Profile if they exist. If they do not, we set them up in your name.

What you see along the way. A draft map before it is final, then a working session where you go down it and cross things off. Owners always cut something we kept and defend something we cut, and they are usually right, because they know which jobs pay and which ones eat an afternoon.

How long, and what happens after. A single location business with a normal service list is usually a week or two of calendar time. Multi location work takes longer, mostly because there are more decisions for you to make. Inside the ongoing program the map is not a one time artifact: it gets revisited as pages ship and as we learn which clusters produced calls, and searches move between tiers on that evidence.

Who needs this, and who should skip it

We talk businesses out of this more often than you would expect, so here is the honest split.

Worth doing now: you are about to spend real money on content or a rebuild and want the target set first; you sell several services or serve several towns, so the number of possible pages is bigger than a napkin holds; traffic does not convert and you cannot tell whether the pages or the audience are the problem; or you inherited a site and need to know what is worth keeping.

Skip it, at least for now.

• You need the phone to ring this month. A keyword map does not produce a lead. It produces a plan whose payoff is months out. Put the money into Google Ads and come back when the pressure is off.
• One service, one small town. Your list is short and you already know most of it. Write it yourself in an afternoon: our free guide to keyword research for local businesses covers the process and the free sources, and it is one of 361 in-depth guides in the learning library at kellywm.com/blog.
• Nobody is going to build against it. The most expensive version of this work is the one that gets delivered, admired, and filed. With no time to produce pages, one page you actually publish beats a plan for thirty.
• The site cannot carry the plan. If pages are not getting indexed, a better target list changes nothing. Start with an SEO audit or the technical SEO work. The free website report card will tell you in seconds whether that is your situation.

Where this sits in the rest of the SEO work

Keyword research is one deliverable inside a larger program, and it comes early, right after the technical foundation, because everything downstream aims at what it produces. Our SEO page covers the order the whole program runs in, what a month of work includes, and what search cannot do no matter who you hire. This page is only the piece that sets the target.

• Who holds these searches now, and why: competitor analysis.
• Turning the list into pages: SEO content writing, one cluster per page.
• Pointing pages you already own at the right cluster: on-page SEO. The first move after a map lands is often repair, not new pages.
• Searches settled on the map instead of a page: local SEO.
• Questions people put to assistants instead of search boxes: AI search. The question inventory feeds that work directly, and the free AI visibility checker shows what those tools say about your business today.
• Knowing whether any of it worked: SEO reporting, measured against the clusters the map named rather than a general feeling of progress.