Search Intent

Search engines work out what a query means before they rank anything. Google describes this first step as understanding the intent behind the query, using language models to match a few typed words to useful content, synonyms and misspellings included^[1]. So they rank the kind of page that satisfies the intent, not the page with the most matching words. A product page struggles for “how to choose a bed frame”, and a blog post struggles for “buy oak platform bed queen”.

A page that ranks on page one and earns few clicks often answers a different question from the one being asked. Google’s guidance on helpful content asks the same thing from the writer’s side: will someone leave the page feeling they learned enough to reach their goal^[2]?

Intent also sets the value of a query. A transactional query with 200 searches a month can be worth more than an informational one with 20,000.

AI assistants read intent too, at the moment they answer. OpenAI’s ChatGPT-User and Perplexity’s Perplexity-User fetch a page when a person’s question needs it^[3][4]. In QueryRev’s study, crawlers of that kind are blocked by name on 3.6% of the top sites read, on average, against 7.9% for training crawlers^[5].

Intent is also the cheapest diagnosis in SEO. Before a page is rewritten, re-linked or merged, reading the queries it already receives often shows that it is the right page for a different question, and that the fix is a new page, not a better version of this one.

Intent is inferred, not reported. The evidence is the wording of the query (how, best, vs, buy, a brand name), the kinds of page that rank for it today, and the features on the results page, such as shopping results or an AI Overview. For your own pages, compare the queries a page earns impressions for with what the page offers. A gap shows up as a click-through rate below what its position should earn.

Label intent per query, not per page. A page usually earns impressions for queries of several intents; the one that brings most of its clicks, or its revenue, is the intent it should serve first.

Watch the results page, not only your page. When the pages that rank for a query change from guides to product listings, the intent has moved, and a page that held its position by serving the old intent will lose it.

Sample data: an invented site, to show the method.

A product page for an oak platform bed sits at position 8. Most of its queries ask about sizes, and the title mentions none; click-through is 1.9%, where this site’s own pages at that position average about 3.4%. Answering the sizing intent on the page is the fix.

The rewrite adds a table of sizes and names them in the title. If click-through moves toward what the position should earn, the mismatch was the cause.