Map search intent before you write

Every keyword carries a why: learn, compare, or buy. Label each one hot, warm, or cold first, so the pages you build match how far along the searcher already is.

Two pages can target equally popular keywords and produce completely different results, because the searches sit at different distances from a decision. Someone asking what a term means is not in the same place as someone typing a competitor's name next to yours, and no amount of writing quality moves a page across that gap.

Intent is the label for that distance. This guide covers how to read it off a keyword list quickly, and what to build for each label once you have it.

Where the intent actually lives

Google does not publish intent. It reveals it, in the results it serves. Whatever format dominates page one for a query is Google's current answer to the question "what do people who type this want", and that answer is derived from behavior rather than from the words in the query.

So the label is read, not guessed. A keyword whose results are all product pages is a buying keyword even if it contains no buying language. A keyword whose results are all long guides is a learning keyword even if it looks commercial. The query text is a hint. The results page is the evidence.

That is also why intent labels go stale. Formats shift, and a label assigned two years ago may describe a results page that no longer exists. Labels attached to a live keyword list need re-reading occasionally, especially before you commit a page to one.

Four intents, three temperatures

The standard taxonomy has four intents, and it is worth keeping the words straight:

  • Informational — the searcher wants an answer or an explanation.
  • Navigational — the searcher wants a specific site or page they already half-know.
  • Commercial — the searcher is comparing options before a decision.
  • Transactional — the searcher intends to act now.

Four labels are right and still too slow for planning a quarter of work, because they do not tell you the order to build in. Folding them into three temperatures does:

  • Hot — transactional and the sharpest commercial terms. The person could act today. "Tool X pricing", "tool X alternative", "best tool for a five-person agency".
  • Warm — commercial comparison at an earlier stage. They have a shortlist forming but no favorite. "Tool X vs tool Y", "how to choose a keyword tool".
  • Cold — informational. They are learning the topic, and may not be a buyer yet at all.

Temperature is a build order, not a quality score. The hottest keywords have the smallest volume and the coldest have the largest, which is exactly why so many content plans end up back-to-front: they are sorted by traffic, and traffic sorts cold first.

Labeling a list without reading a thousand SERPs

You can label most of a list from the words, and reserve the SERP for the ones that resist.

Start with the signals in the query itself. Terms with a product category, a competitor name, a "vs", a "pricing", an "alternative", or a modifier like "best" or "cheap" are hot or warm. Terms starting with "what", "how", or "why" are cold unless the rest of the query says otherwise. Navigational terms contain a brand, and the only question is whether the brand is yours.

Then run the doubtful ones past a results page. In Doverity's keyword research the intent label comes attached to each keyword, which handles the bulk of the list, and get_serp_results settles the remainder by showing you what format is actually ranking. If a keyword you labeled cold returns pricing pages, the label was wrong and the SERP just told you so.

Expect to relabel a meaningful share. A list where nothing changes after the SERP check is a list that was labeled from the results page in the first place, which is fine, but unusual.

What each temperature gets built as

The label is only useful because it decides the page.

Hot keywords want a page that can close. Product pages, pricing pages, comparison pages, and alternatives pages. The person typing wants a reason to pick you or a reason to leave, so the page needs the specifics: what it does, who it is for, what it costs, and where it beats the thing they were already considering. Burying a hot keyword inside a 2,000-word explainer is the most common waste in content planning.

Warm keywords want a page that helps them compare. Honest breakdowns, use-case pages, and decision guides that name tradeoffs. These convert later than hot pages and convert better than cold ones, and they are the pages most likely to be cited when somebody else's buyer asks around.

Cold keywords want a page that earns trust and passes it on. The direct return is small and slow. The real job is to be the page a warm searcher remembers, and to link onward to the warm and hot pages so the visit has somewhere to go. A cold page with no internal path to a decision is an exit.

That linking is the practical reason to label before writing rather than after. A labeled list gives you a site structure where every page knows which temperature it is serving and which page it hands to.

Do it with an agent

Connect the Doverity MCP and the labeling runs over the whole list instead of a handful of rows. Use this prompt, or read the keyword research skill, which follows the same order.

Here is my keyword list: [paste]. If I give no list, run
research_keywords for "[topic]" first.

1. Give every keyword its intent: informational, navigational,
commercial, or transactional. Fold those into hot, warm, and cold.
2. For every keyword you are unsure about, call get_serp_results and
tell me what format is ranking: product pages, comparisons, guides,
or videos. Change the label if the SERP disagrees with the words.
3. Sort the table by temperature, then by how closely the query
matches what my project sells.
4. For each hot and warm keyword, name the page type it needs and say
whether a page of mine already covers it, using
get_search_console_performance to check.

The last column — page already exists, or page missing — is the plan. Everything else in that table is supporting detail.

Search intent FAQ

Can one page target two intents?

Only when the SERPs agree. If the same results rank for both queries, one page can hold both. If one query returns pricing pages and the other returns guides, Google is treating them as different jobs, and a page built for one will sit out the other. Check the overlap before merging, and split when the formats differ.

What are buyer-intent keywords?

The terms people use when a decision is close: a product category, a competitor's name, "pricing", "alternative", "vs", "demo", or "best" plus a narrow segment. They are searched less than the surrounding educational terms and they convert far better, because the person typing them has already decided to solve the problem and is only choosing how.

How do I check the intent of a keyword?

Look at the results page. Whatever dominates it is the intent Google has assigned, and it beats any label from the words alone. A keyword tool will pre-label most of a list for you, which is worth doing for speed; keep the manual check for the terms you intend to build pages against, since those are the ones where a wrong label costs a page.

Does intent matter more than volume?

For a business with limited capacity, yes. A high-volume keyword with the wrong intent produces traffic that never turns into anything, and a low-volume keyword with buying intent produces customers. Sort by intent first and use volume to choose within a temperature, not across them.

Run this strategy with an agent

Every guide includes a prompt you can paste to an agent. With the Doverity MCP connected, it reads your Search Console, rank data, and backlink profile while it works. Free credits at sign-up, or self-host for nothing.

Stay in the loop

Product updates, new features, and the occasional behind-the-scenes.