Keyword Research for People Who Hate Keyword Research

Keyword research is the part of SEO most likely to make a normal person give up. The tutorials all describe the same ritual: sign up for a tool, export thousands of keywords into a spreadsheet, study columns of search volume and difficulty scores, build a matrix. If you find this repellent, I want to tell you your instincts are fine. The ritual is mostly cargo cult. The thing underneath it is not only simple, you’re probably better positioned to do it than the people selling the tools.

Because here’s what keyword research actually is: figuring out what words your customers use. Not what words you use. What they use. That’s the entire discipline. Everything else — the volumes, the difficulty scores, the spreadsheets — is apparatus built around that one question, and the apparatus has a way of crowding out the question.

The reason the question matters is that businesses reliably drift into their own dialect. You say “pre-owned certified vehicles”; your customer types “used cars.” You say “revenue operations platform”; your customer types “how to stop deals falling through the cracks.” You say “periodontal therapy”; your customer types “bleeding gums.” This drift is natural — expertise changes your vocabulary — but search engines match words, and every gap between your dialect and your customer’s dialect is a search you silently sit out. The most common keyword mistake isn’t targeting the wrong keywords. It’s writing an entire website in a language your buyers don’t speak.

So the first and best research method is listening. You already have the data; it arrives daily and you’ve been discarding it. What do people say on the phone in the first thirty seconds? What words are in the emails asking whether you handle such-and-such? When a customer describes their problem before you’ve taught them your terminology — that’s the keyword. Write these down verbatim for two weeks. Resist the urge to translate into your professional vocabulary, because the translation is exactly what you’re trying to catch. A landscaper who does this discovers customers never say “hardscaping.” They say “patio.” That single observation is worth more than a thousand-row export, because it changes what you put in your titles and headings, which is where matching actually happens.

The second method is watching search engines finish your sentences. Type the beginning of a customer-style phrase into Google and read the autocomplete suggestions. Those aren’t guesses; they’re drawn from what people actually search. Then run the search and look at the “people also ask” questions and the related searches at the bottom. In ten minutes you can map the real vocabulary of a topic — which phrasings exist, which questions cluster around them — using the search engine itself as the instrument. It won’t give you volume numbers. You don’t need them. You’re not arbitraging thousands of keywords; you’re trying to describe a handful of pages in the words their readers use. For that, knowing “people phrase it this way, not that way” is the whole prize.

The third method is looking at who currently ranks. Search the phrase you care about and study the first page as evidence of what Google believes the phrase means. If you search a phrase intending to sell something and the results are all Wikipedia-style explainers, you’ve learned the phrase signals research, not buying — the searcher wants to understand, and a product page would be an awkward answer. This is what the tools call search intent, and you can read it directly, no subscription required. The results page is Google showing you its homework.

Then you apply what you found, and the application is almost anticlimactic. You take the customer’s phrasing and put it in the places that declare what a page is about: the title tag, the H1 heading, the meta description, the first paragraph. Not stuffed and repeated — used, the way you’d naturally title a document about that subject for that reader. When I look at audits from GazeSite, the tool I built for this, the findings on struggling pages are rarely “targeted a keyword with the wrong difficulty score.” They’re “No H1 heading found” and “Missing meta description” — pages that did no declaring at all, in anyone’s dialect. The sophisticated failure the tools guard against is rare. The basic failure is everywhere.

I should be fair to the tools: at scale, they earn their keep. If you run a fifty-thousand-page site or a content team shipping daily, volume estimates and rank tracking genuinely inform decisions. But that describes almost none of the people made miserable by keyword-research tutorials. For a site with ten or thirty pages that matter, the listening method isn’t the beginner’s compromise. It’s the appropriate tool for the job, and it produces something exports can’t: sentences in your customers’ own voice, which turn out to be exactly what you needed to write anyway. The research and the writing converge. That’s how you know you’re doing the real thing and not the ritual.