Keyword research and strategy

The four kinds of search intent, and which ones pay

Every search sits in one of four buckets, and knowing which one you are looking at decides everything that follows. Informational searches want to learn something. Navigational searches want a specific site or brand. Commercial searches are comparing options before buying. Transactional searches are ready to act now.

Commercial and transactional searches produce enquiries. Informational searches produce traffic, which sometimes becomes enquiries later and often does not. Neither is wrong to target, but they serve different purposes and a plan that mixes them without saying which is which will disappoint somebody.

You cannot always tell intent from the words. The reliable method is to search the term and read what Google is already showing, because the results are the accumulated evidence of what satisfied millions of previous searchers. If the page one results are all service pages, the intent is commercial no matter how informational the phrasing sounds.

• Informational: builds audience and topical authority, converts slowly
• Commercial: comparison and evaluation searches, the highest value for most services
• Transactional: ready to buy or enquire, lowest volume and highest worth
• Navigational: your own brand terms, which you should own completely

Why search volume numbers are less reliable than they look

The monthly volume figures in every keyword tool are estimates, not counts. They come from Google Ads data that is bucketed and rounded, then modelled by each tool with its own assumptions. Two reputable tools will routinely disagree by a factor of three on the same term, and both can be wrong.

Close variants are also grouped together in ways that hide real demand. Singular and plural forms, word order changes and near-synonyms often report identical volumes because they are treated as one bucket. That means the long tail is systematically underestimated, and terms showing zero volume can still bring steady traffic.

We use volume for relative comparison rather than forecasting. It is useful for deciding whether term A is bigger than term B. It is not useful for promising a specific number of visits. Where we want reality rather than estimates, we use your own Search Console impression data, which is a count of what actually happened.

Judging whether you can actually rank

Keyword difficulty scores in tools are almost entirely based on the backlinks pointing at the pages currently ranking. That is one input and it misses several that matter more in practice. A better assessment asks what kind of sites are ranking, how well those specific pages answer the query, and whether there is an obvious weakness you can beat.

The most useful signal is heterogeneity. If page one is entirely enormous national brands and marketplaces, a small site has no realistic route in regardless of what the difficulty score says. If page one contains a couple of thin, out-of-date pages or forum threads, the query is winnable even with a high score, because Google is visibly short of good answers.

Your own site profile changes the answer too. A site with three years of history and real authority in a subject can target terms that a site launched last month cannot. Winnability is a relationship between the query and you, not a property of the query.

• Who is ranking: national brands, marketplaces, or sites your size
• How good the ranking pages actually are when you read them
• Whether results features leave meaningful organic clicks available
• How much authority your site already has on the surrounding topic

Finding the questions tools do not report

The most valuable terms for service businesses are often long, specific and barely register in volume tools. They are also the ones that convert, because somebody typing an eleven-word question describing their exact situation is much closer to hiring than somebody typing a two-word category term.

Tools will not give you most of these. The sources that will are the places your buyers already talk. Your own sales calls and the questions that come up in them. Support tickets and email threads. The People Also Ask chains that expand endlessly around your main terms. Community threads and subreddits in your field. Review sites, where people describe what they wanted and did not get.

This research also produces the phrasing people genuinely use, which is usually different from industry vocabulary. Matching that phrasing matters more now that people ask conversational questions of AI assistants rather than typing keyword fragments.

Mapping keywords to pages before writing anything

A keyword list is not a plan. The deliverable that matters is a map: every target term assigned to exactly one page, with that page named, its intent stated, and its purpose in the funnel clear. Without that map you get two predictable failures.

The first is cannibalisation, where several pages chase the same query and none of them wins. The second is orphan research, where a large list is produced, admired and never acted on, because nobody decided which page each term belonged to.

Grouping matters as much as assignment. Terms that describe the same need in different words belong on one page, because Google understands synonyms and a single strong page will rank for all of them. Terms that look similar but represent different needs belong on different pages. The test is the results page: if the same URLs rank for both terms, one page can serve both.

• One primary term and intent per page, with supporting variants grouped under it
• Existing pages assigned before new ones are proposed
• Gaps identified as specific pages to build, not vague topics
• Priority set by intent value and winnability, not by volume

Seasonality, trends and when to publish

Some demand is steady and some is not, and treating a seasonal term like an evergreen one wastes months. Publishing a page in the week that demand peaks is too late, because a new page needs time to be crawled, evaluated and to accumulate any signals at all.

For seasonal terms we work backwards. Identify when demand begins to build from historical trend data, then publish and start promoting well ahead of that. Keep the same URL year after year, because a page in its third season carries accumulated authority that a freshly created page cannot match.

Trend direction matters as much as current volume. A term with modest volume that has doubled over two years is usually a better investment than a larger term in steady decline, because you are building for the demand that will exist when the page matures.