Research & briefs
The four search intent types, and how to tell them apart on a real SERP
Intent is not a property of the words in a query. It is what the results page shows people accepted, and that is the only place you can read it reliably.
7 min read
Search intent is the goal behind a query: learn something, reach a specific place, weigh options, or complete an action. The four-way split has been standard in information retrieval since Broder's 2002 taxonomy, and the commercial-investigation category was added later because buying research behaves differently from both learning and buying.
The categories are useful, but the common way of applying them is not. Classifying by phrasing — "how to" means informational, "buy" means transactional — fails on exactly the queries where the decision matters, because the same words carry different intent in different markets.
This article covers what separates the four types, the SERP evidence that identifies each, the mixed and shifting cases that break the model, and how to diagnose a page that is losing to intent mismatch rather than to competition.
The four types, side by side
The distinction that matters operationally is what the searcher wants to do next, not what they typed.
| Intent | The goal | Typical SERP | Page you should build |
|---|---|---|---|
| Informational | Understand something | Guides, definitions, PAA, video, sometimes an AI overview | Article, tutorial, reference — answer-first |
| Navigational | Reach a specific site or page | One brand dominating, sitelinks, brand knowledge panel | Nothing, unless it is your brand |
| Commercial investigation | Compare options before deciding | Listicles, reviews, comparisons, mixed vendor pages, ads | Comparison, alternatives, or an honest review |
| Transactional | Do the thing — buy, sign up, download | Product and pricing pages, shopping units, heavy ads | Product, pricing, tool or sign-up page |
Why phrasing is a hint and the SERP is the verdict
Modifier heuristics are fast and roughly two-thirds right, which is worse than it sounds — the third they get wrong contains most of the pages worth building.
| Query | What the wording suggests | What the SERP usually shows |
|---|---|---|
| seo content brief template | Informational | Mostly informational, but the ranking pages all contain a downloadable or copyable artefact — the intent includes getting a thing |
| best crm | Commercial investigation | Confirmed: listicles and reviews, with vendor pages rarely above position five |
| notion | Ambiguous | Navigational: one brand with sitelinks. No third-party page is winnable here |
| how to write a meta description | Informational | Confirmed, but with a snippet and an overview taking the definitional part |
| seo tools | Commercial investigation | Mixed: listicles and vendor pages interleaved — an unresolved SERP where either format can rank |
The tells for each type
Informational
One caveat worth carrying: an informational SERP with an overview at the top has less click value than it looks. The page still needs a reason to be opened after the summary has been read.
- People Also Ask sits high on the page.
- A featured snippet or AI overview answers the core question.
- Publishers, documentation and educational sites outnumber vendors.
- Few or no ads — advertisers do not bid where nobody buys.
Navigational
Treat these as unavailable. The exception is a genuine comparison or alternatives page for a well-known product, which serves a different query — "brand alternatives" — and should be built for that query, not for the brand name.
- A single brand occupies the first result with sitelinks below it.
- A knowledge panel for that brand appears on the right.
- Positions two to five are the same brand's subpages, its app listing or its social profiles.
Commercial investigation
This is the largest category in most content plans, and the one where an independent publisher can still beat a vendor, because readers actively distrust a comparison written by one of the options.
- Listicles with numbers in the title dominate the first page.
- Review sites and affiliate publishers appear alongside vendors.
- "Best", "vs", "alternatives" and "review" appear in the titles of ranking pages, even when they are not in your query.
- Ads are present but not overwhelming.
Transactional
Free tools sit here more often than people expect. A query like "word counter" is transactional in behaviour — the searcher wants to perform an action, not read about it — and an article will not rank against pages that just do it.
- Product, pricing and category pages fill the organic results.
- Shopping or local packs occupy the top of the page.
- A dense ad block sits above the first organic result.
- Ranking pages are short, with prices, specifications or a sign-up form rather than prose.
Mixed, shifting and fractured intent
- Mixed SERPs — Several page types share the first page because searchers want different things from the same words. Pick the dominant sub-intent, serve it properly, and mention the others in a paragraph with a link rather than trying to be all of them.
- Shifting intent — Intent moves. A product category becomes a research topic once the market matures; a news-driven query is transactional for a season and informational afterwards. Re-check the SERP before any major rewrite.
- Fractured by device or location — The same query returns a local pack on mobile and an article-heavy SERP on desktop. Check both when the topic has any geographic dimension.
- Intent inside a single page — A long informational page often carries a transactional tail — the reader who now wants the tool. That is a CTA at the end, not a second page.
Diagnosing an intent mismatch
Mismatch has a recognisable signature and is not fixable by writing more. These are the symptoms and what each one implies.
| Symptom | Likely cause | Fix |
|---|---|---|
| Ranks 15-40 for months, never breaks into the top ten | Format does not match what the SERP rewards | Rebuild the page type — article to comparison, prose to tool |
| Good position, poor click-through rate | The title promises a different page type than the searcher wants | Rewrite the title to signal the format, or accept the query is not yours |
| Good click-through rate, immediate exits | The page answers a different question than the query asked | Move the real answer to the opening; cut the preamble |
| Impressions on queries you never targeted | The page is being interpreted as serving another intent | Follow the evidence — that intent may be the better page |
| Traffic fell without a position change | A feature or overview now answers the query above you | Add what a summary cannot provide: procedure, data, tooling |
Using intent as a planning filter
- Assign an intent to every query in the plan before assigning a page. Two queries with different intents are always two pages.
- Two queries with the same intent and substantially the same SERP are one page — this is the check that prevents cannibalisation inside a cluster.
- Drop navigational queries for brands you do not own; there is no version of that page that wins.
- Sequence commercial-investigation pages ahead of transactional ones when you are new: comparisons are winnable with content, product pages usually are not.
- Record the intent verdict in the brief, because it is the field every later decision — format, length, CTA — depends on.
Questions people ask
How many types of search intent are there?
- Four in common practice: informational, navigational, commercial investigation and transactional. The original academic taxonomy had three — informational, navigational, transactional — and commercial investigation was split out because comparison behaviour differs from both learning and buying.
Can one page serve two intents?
- Partly. A page can serve a dominant intent and address a secondary one in a section with a link. A page that tries to be both a guide and a product page is usually beaten by pages that committed to one.
Is search intent a ranking factor?
- Not as a named factor. It is a description of what the ranking systems have already selected for. Matching it is how you become eligible; missing it caps the position regardless of how good the page is.
How do I identify intent for a query with no search volume?
- The same way: search it. A SERP exists for queries with no reported volume, and the page types on it are the evidence. Volume tools go quiet long before the results page does.
References
- Broder, A. — A taxonomy of web search (SIGIR Forum, 2002) — The original informational / navigational / transactional classification the model is built on.
- Google Search Quality Rater Guidelines — Google's own framing of user intent and "Needs Met" as raters are instructed to assess it.
- Google Search Central — Creating helpful, reliable, people-first content — The people-first questions that describe intent satisfaction without naming it as a factor.
- Google Search Console — Performance report — The impression, position and CTR data used to diagnose the mismatch signatures above.
Keep reading
- The pass that produces the intent verdict — How to read page types, features and diversity from a live results page in ten minutes.
- Where the verdict gets recorded — The brief field that every later format and length decision depends on.
- Auditing pages that lost on intent — Deciding whether a mismatched page should be rebuilt, merged or dropped.
- Intent classification inside the draft process — What wordiqo detects during research and checks against before returning an article.