# Search Intent Types: The Real Origin and How to Use Them

> Search intent, traced to its real 2002 origin and 2025 AI-chat data, plus a decision table matching page type and funnel stage to each of its 6 types.

- URL: https://missiongrowth.io/blog/search-intent
- Published: 2026-08-03 · Updated: 2026-09-23
- Author: Ömer Furkan Aktaş, Founder, Mission Growth
- Publisher: Mission Growth. Company facts: https://missiongrowth.io/llms.txt

Search intent is usually reduced to four labels, informational, navigational, commercial, transactional, repeated without a source behind them. The full picture traces to a research paper that named three categories, not four, and has been extended twice since. A repeatable classification method and a decision table turn that intent into a page type and a funnel stage.

## What search intent is (and where the “4 types” model actually comes from)

Search intent is the underlying goal behind a search query, sometimes called user intent.

What is user intent asking, if not the same question in different words? Both terms describe the same taxonomy, and that taxonomy has a real, citable origin: Andrei Broder's paper, "A Taxonomy of Web Search," published in SIGIR Forum.

Broder defined three categories, not four: navigational (find a specific site), informational (learn something) and transactional (complete an action, usually a purchase). The four-type version gets stated as settled fact everywhere, with no source attached and no note that a fourth category was added later.

Tracing the types of search intent back to Broder's original three shows exactly which parts of today's model are original research and which are a later industry addition. Broder backed his three categories with two measurements, not one: an AltaVista survey and a separate query-log analysis.

The survey collected 3,190 valid responses between June and November 2001, splitting navigational at 24.5%, informational at an estimated 39% and transactional at over 22%, estimated around 36%. The log analysis found a different mix: 400 queries drawn from a random 1,000-query sample split navigational 20%, informational 48%, transactional 30%.

Two measurement methods, two different splits, from the same paper. A follow-up study by Rose and Levinson revisited the taxonomy and found navigational search less common than assumed. Their paper proposes a resource-seeking goal that replaces Broder's transactional category at the top of their hierarchy, rather than adding a fourth option beside it.

Today's commercial intent is a later industry addition to Broder's original three, a different move from the one Rose and Levinson made. Treating the two as the same correction is a common mix-up.

::figure{src="/blog/figures/search-intent-1.svg" alt="Timeline of the search intent framework: Broder's taxonomy, Rose and Levinson's refinement, industry convention, and 2025 NBER data" caption="The search intent framework traces back to a narrower original taxonomy, and forward to 2025 generative-intent data" width="720" height="239"}

## The six types of search intent, with real examples

Search intent breaks into six practical categories in current industry use: informational, navigational, commercial, transactional, local and generative.

The table below gives search intent examples for each type, paired with the SERP signal that confirms it. Most real queries carry more than one type at once.

::dataset{key="search-intent-six-types" name="Six search intent types, definitions and dominant SERP signals"}

| Type | What the searcher wants | Example keyword | Dominant SERP signal |
|---|---|---|---|
| Informational | To learn or understand something | what is search intent | Featured snippets, People Also Ask, definition-style results |
| Navigational | A specific site or brand | ahrefs login | One brand dominates, sitelinks appear |
| Commercial | To compare options before buying | best rank tracker | Listicles, comparison tables, review sites |
| Transactional | To buy or act right now | buy seo software subscription | Product pages, pricing pages, shopping results |
| Local | A nearby physical result | seo agency near me | Map pack, Google Business Profile listings |
| Generative | A synthesized answer, not a page to click | why is chatgpt giving a different seo answer than google | A conversational answer inside a chat interface, no ranked list |

Commercial and generative are the two industry additions that came after Broder's original three; local followed once "near me" search matured. None of these extensions overturn Broder's taxonomy. They describe territory his paper never tried to cover.

The commercial intent vs transactional intent line is the one practitioners misapply most, because a single keyword often carries both. Take "best CRM software" and run the SERP check: the top 10 mixes comparison articles, a commercial signal, with vendor pages offering a free trial or demo, a transactional signal.

Count what dominates. Comparison content outnumbers vendor pages in that SERP, so commercial is the primary intent, with transactional running a close second. That reading decides the page to build: a comparison, but with a low-friction trial link built into it rather than a contact form, so it still serves the transactional-leaning share of that traffic.

## How to determine the dominant intent behind any keyword

The SERP is the ground truth for a keyword's intent, and three checks are how to determine search intent from it.

Run through them in order:

1. **Check the SERP's result types first.** Blog posts signal one intent; a map pack or shopping results signal another. This composition outranks the keyword's own wording, since Google has already classified the query for you.
2. **Apply Ahrefs' "3 Cs."** Content type (blog post, product page or category page), content format (listicle, how-to, comparison or review) and content angle (beginner-focused, price-focused or brand-specific). Ahrefs frames those three questions as a repeatable way to turn "check the SERP" into an actual checklist.
3. **Cross-check an automated classifier, then treat its label as a hypothesis to confirm.** Keyword tools that tag intent automatically often disagree with a manual SERP read.

A Google patent on context-based intent prediction (US20180336200A1) describes predicting a query's intent from the device, location and time it was searched, not only the query text, since the same keyword can point to a different intent depending on where and when it's typed. Confirm any tool's label against the actual SERP before a decision rests on it.

That same three-step check does not scale cleanly to a keyword list running into the thousands. Group the list by intent-signaling modifiers first, best, vs, buy, near me, how to, what is, then spot-check a sample of each group's SERP rather than running the full check on every row.

Reading the pages that already rank is worth the extra step beyond checking their format. Check how deep they go and whether they answer the question behind the keyword. A modifier signals the shape of the page to build; the competing content signals how much page to build.

## Matching content and funnel stage to intent (a decision table)

Each of the six intent types maps to one dominant page format and one funnel stage.

Building the wrong page type for a query's intent is a common reason an otherwise well-optimized page still doesn't rank. That question is different from whether a keyword is worth pursuing at all.

Once the intent type here is known, [how to rank higher on google](https://missiongrowth.io/blog/how-to-rank-higher-on-google) covers the separate decision of whether the top 10's coverage gap makes a keyword worth targeting now, reshaping for, or skipping, a test this table doesn't replace.

::figure{src="/blog/figures/search-intent-2.svg" alt="Table matrix mapping each of six intent types to its dominant SERP signal, the page type to build, and the funnel stage it sits at" caption="One lookup: intent type decides both the page type to build and the funnel stage" width="720" height="423"}

Take the informational type: an explainer or how-to guide, at the awareness stage. Transactional belongs on a product or pricing page, at the decision stage. Navigational points at the brand or product page the searcher already has in mind, which puts them somewhere between aware and existing customer. Local wants a location page tied to a Google Business Profile, from a buyer close enough to walk in.

Commercial intent carries outsized weight for conversion. A searcher this close to a purchase decision is comparing your page against a competitor's in the same browsing session, and converts or doesn't on which page makes the stronger case.

So a mismatched page type costs more here than anywhere else. It doesn't just lose a click, it hands the comparison to whoever else showed up in that SERP.

Generative intent is the row worth pausing on, since it doesn't fit the click-through model the other five assume. A page built for a generative query is competing to be the passage an AI answer quotes, not the link a searcher clicks, so structure and quotable statements matter more than a compelling headline.

Optimizing a transactional page means stripping friction, not adding persuasion: a clear price or a fast quote path, one dominant call to action, and proof, reviews, guarantees, live chat, placed where the buyer is already ready to act.

A commercial page runs the opposite move. It can name a price range to keep the comparison honest, but a full pricing table or a hard buy button belongs on the transactional page it links to, not on the page still doing the comparing.

## Generative search intent: what 2025 ChatGPT usage data actually shows

A study of ChatGPT conversations found people split roughly 49% asking for information, 40% asking the model to do a task and 11% simply expressing themselves, a sourced, dated picture of generative intent.

The study, "How People Use ChatGPT," comes from Chatterji and six co-authors at the National Bureau of Economic Research. It randomly sampled roughly 1.1M ChatGPT conversations logged between May 2024 and June 2025, and sorted them into three buckets: Asking, Doing and Expressing.

::dataset{key="search-intent-chatgpt-usage" name="NBER classification of ChatGPT messages by usage type, May 2024-June 2025 sample"}

| Usage type | Share of messages | What it covers |
|---|---|---|
| Asking | 49% | Seeking information or an answer |
| Doing | 40% | Asking the model to complete a task |
| Expressing | 11% | Venting, roleplay or other non-search use |

::figure{src="/blog/figures/search-intent-3.svg" alt="Bar chart showing NBER's classification of ChatGPT messages: 49% Asking, 40% Doing, 11% Expressing" caption="What ChatGPT users are actually doing: Asking, Doing or Expressing" width="720" height="363"}

Asking maps loosely onto informational intent, and Doing maps loosely onto transactional intent, in the sense both involve the model completing something rather than just answering a question. Expressing has no real equivalent in Broder's original taxonomy, since it describes conversational use that never resembles a search query at all.

For the separate question of how much prompt volume this represents across AI platforms, [prompt volume data](https://missiongrowth.io/blog/prompt-volume) covers that angle using the same NBER dataset's message-volume figures, a different cut of the same underlying sample.

## Why intent match still decides both rankings and AI Overview exposure

An AI Overview doesn't retire search intent, it adds a second gate next to the ranking one.

Google's ranking systems reward pages that match a query's intent, and AI Overviews trigger far more on informational queries than on commercial or transactional ones. Misreading intent now risks losing a ranking and a citation in the same query.

That second half is no reason to treat search intent as settled once an AI Overview enters a SERP. [AI overviews impact on seo](https://missiongrowth.io/blog/what-ai-overviews-mean-for-seo) covers how unevenly that exposure lands across intent types. The mechanism deciding whether a page gets considered for an AI Overview is the same intent match this guide has walked through, applied to a new surface.

Search intent traces to real research, narrower at its origin than most people assume, extended twice since, and now gaining a third layer as more search happens inside a chat window instead of a results page.

The corrected model runs six search intent types, one origin story, and one lookup from intent to page type and funnel stage.

Pick one keyword already ranking on the wrong page type. Run it through the decision table above, check which row it actually belongs to, and rebuild that page around the format and funnel stage the row specifies.

## FAQ

### "Search intent" = buzzword?

No. It traces to specific information-retrieval research, Andrei Broder's original taxonomy, even though the everyday four-type version in circulation has drifted from that original research. A term with a named paper behind it and a 2025 dataset extending it earns more credibility than the buzzword complaint suggests.

### How accurate is automated or AI search-intent classification?

Automated classifiers disagree with each other and with a manual SERP check often enough that a label deserves a spot-check before it drives a decision. Treat any tool's output as a starting hypothesis, and confirm it on the actual SERP.

### Should a brand-new site target informational-intent keywords first?

Not because of intent itself; the decision table above applies the same way regardless of how old a site is. Many new sites start with informational content anyway for a separate, practical reason: commercial and transactional pages need trust signals, reviews, case studies, backlinks, that a new site usually hasn't earned yet.

### What metrics best show whether content matches search intent?

SERP-retention signals are closer proxies than rankings alone: whether the searcher clicks back to the results, how far they scroll, whether they re-search with a refined query. A page can rank and still fail the intent test that determines whether it keeps that ranking.

### Are low-volume commercial-intent keywords worth targeting?

Often yes. Thin search volume on a specific, bottom-funnel query usually signals a small buying committee that is still worth reaching. [b2b keyword research](https://missiongrowth.io/blog/b2b-seo) covers the full method for finding and prioritizing exactly this kind of keyword.