How to do keyword research, intent first

Most keyword research goes wrong in the first five minutes, when someone opens a tool, sorts by search volume, and starts writing down the biggest numbers. It feels productive. It produces a list of terms you will never rank for, aimed at people who were never going to buy. The volume column is the most seductive and least useful thing on the screen.

Keyword research is not a hunt for big numbers. It is the work of understanding what people type, why they type it, and which of those searches your business can actually satisfy. Done properly, it is closer to reading customers than to operating software.

Why intent beats volume

A keyword's value is not its search volume — it is the volume multiplied by how ready the searcher is to act, multiplied by your realistic odds of ranking. Two of those three factors have nothing to do with the big number in the tool.

Take an illustrative pair. "What is SEO" might return 12,000 searches a month; "SEO agency Singapore pricing" might return 90. The first is a student, a curious founder, someone three years from ever hiring anyone. The second is a person with a budget, a shortlist, and a decision to make this quarter. If you sell SEO, the 90-search term is worth more than the 12,000-search one — and it is usually far easier to rank for, because fewer people bother to target it. Chasing volume gets that exactly backwards.

This is the core of an SEO programme that produces leads rather than traffic reports: you weight terms by intent and winnability first, and treat raw volume as a tie-breaker, not the headline.

The four intent types

Every query sits somewhere on a spectrum from "just looking" to "ready to buy". Four buckets are enough to work with:

• Informational — the searcher wants to learn. "How does keyword research work", "what is search intent". They are not ready to buy; they are ready to read. You earn trust here, not conversions.
• Commercial investigation — the searcher is comparing before committing. "Best SEO agency Singapore", "Ahrefs vs Semrush", "SEO agency reviews". This is the most under-valued bucket: real buying intent, but still open to persuasion.
• Transactional — the searcher wants to act now. "SEO agency Singapore pricing", "hire SEO consultant", "free SEO audit". Lower volume, highest commercial value.
• Navigational — the searcher wants a specific place. "Media On Tap contact", a brand name. You rank for your own; you rarely win someone else's.

The mistake is treating these as interchangeable. An informational query answered with a hard sales pitch fails; a transactional query answered with a 3,000-word explainer sends the buyer to a competitor's checkout. Match the page to the bucket.

Read the SERP — Google has already decided

You do not have to guess intent, because Google has spent billions inferring it and shows you the answer for free. Search your target term and read the results page as evidence. The pages that already rank tell you what Google believes the searcher wants.

If the first page is all listicles and guides, the intent is informational — a product page will not break in. If it is dominated by service pages and pricing, the intent is transactional, and your explainer article has no place there. Watch the SERP features too: a shopping carousel signals commercial intent, a "People also ask" box shows the adjacent questions, and a local map pack means Google reads the query as location-driven. When the results mix page types, the intent is genuinely split, and that is worth knowing before you write a word.

This one habit — reading the SERP before committing to a keyword — prevents more wasted content than any tool. If everything ranking looks nothing like the page you intended to build, the term is not for that page.

One cluster, one page

Once you understand intent, group your keywords by the job they represent, not by how similar the words look. "SEO cost Singapore", "how much does SEO cost", and "SEO pricing Singapore" are three phrasings of one question — they belong to one page. "What is SEO" is a different job and belongs to a different page.

Map one keyword cluster to one page, and give that page the whole cluster. The failure mode is cannibalisation: you publish three thin pages all targeting the same intent, they compete with each other, and Google, unsure which to trust, ranks none of them well. You have split your own authority three ways. One strong page targeting a cluster almost always beats three weak ones chasing the same searchers — a point we make in the complete guide to SEO as well.

Build from long-tail towards head terms

The instinct is to target the biggest, most competitive "head" term first — "SEO Singapore", say. On a young or mid-sized site, that term is defended by domains with years of authority, and you will sit on page four indefinitely.

Start instead with the winnable long-tail: specific, lower-volume, higher-intent phrases where the competition is thinner. "SEO for aesthetic clinics Singapore" is far more winnable than "SEO Singapore", converts better because it is specific, and — this is the part people miss — ranking for the specific terms builds the topical authority that eventually lets you compete for the broad ones. You earn the head term by first owning the cluster of long-tail terms beneath it. Trying to skip that step is why so many sites target their most ambitious keyword and rank for nothing at all.