Search intent
What the searcher actually wants behind the words they typed: to learn, to reach a specific site, to compare options, or to buy right now.
See it
What it is
Four usual buckets: informational ('how does dithering work'), navigational ('figma login'), commercial investigation ('best note apps'), and transactional ('buy standing desk'). Matching intent is the cheapest ranking lever there is, because a page in the wrong format cannot win regardless of quality. A product page will not beat tutorials for 'how to', and a 3,000 word essay will not beat a pricing page for 'buy'.
You do not guess intent, you read it off the SERP. Whatever mix of formats ranks today is the engine's own answer about what the query means, assembled from many signals (click behaviour may be one of them, but nobody outside the engine knows the recipe). Read the mix literally: three tutorials, three product pages, and a video means the query genuinely carries more than one intent, and mixed SERPs are normal. Query modifiers help too: 'how', 'what is', 'vs', 'best', 'near me', 'buy', 'template', 'pricing'.
Two things people get wrong. Intent drifts: a query can turn commercial in a season, and a page that ranked happily for two years quietly loses to a new format. And the one-page-one-intent rule gets applied far too hard. A single good page often satisfies several compatible intents at once. Split only when the intents need materially different experiences: when you catch yourself bolting a full checkout flow onto a definition article, that is two pages, not one.
Ask AI for it
Classify the search intent of each query below as informational, navigational, commercial investigation, or transactional, using the live SERP result types and the query modifiers as evidence. For each query give me the page type that should win (tutorial, comparison list, product page, pricing page, free tool, glossary entry), the format the top results share, and the single next action the searcher wants to take. Where the top results are mixed, say so rather than forcing one label. Flag every query where my existing page type does not match the intent, and only recommend splitting a page when the intents would need materially different experiences.