The search intent is a classification concept from the Search engine optimization and content marketing. It captures the intent behind a single search query. So, why someone is searching, not what they type.
In SEO practice, search intent is generally divided into five types
- informational
- navigational
- commercial
- transactional
- local
Google itself also distinguishes between Know, Do, Website and Visit in Person.
Regardless of the subdivision, the purpose behind the search intent is always to align the content format, page structure, and conversion path with the actual intention. .
That's why it's relevant for classic search engine optimization, content strategy, and optimizing for generative search systems like ChatGPT, Perplexity, or Google AI Overviews.
Three terms or concepts are regularly confused here:
- The Keyword The typed-in string itself. However, the same string can have different search intentions. Example: „Beetle“. Is this about the insect or the car?
- The User intention whereas the overarching motivation is the entire Research or shopping trip; the search intent is the respective snippet of that per request.
- Buyer Intent Data again, behavioral signals are at the company level. They show, which companies research, no what intention is behind a query.
The Five Types of Search Intent at a Glance
SEO practice today distinguishes five types. Each can be identified by typical keyword patterns and requires a different content format.
Informational: The user wants to know something. Recognizable by W-questions and knowledge keywords: „What is...“, „how does it work...“, „how it works“, „difference between...“. Example: „Servo drive how it works“. Suitable format: Glossary, Guide, How-to, Explainer video.
Navigational: The user wants to go to a specific page. Recognizable by brand, product, or URL names: „webraketen“ or „web.de Login“. The user already knows the destination and uses search only as a shortcut. Suitable format: Brand homepage, product overview, clearly named subpage.
Commercial Investigation: The user compares before buying. Recognizable by comparison and evaluation keywords: „best,“ „test,“ „comparison,“ „alternative to...,“ „selection criteria,“ „provider for....“. Example: „Servo drive selection criteria.“ Suitable format: comparison table, solution page, selection guide, case study.
Transactional: The user wants to take action. Recognizable by action keywords and specific product names: „buy,“ „order,“ „download,“ „request quote,“ but also pure part numbers and SKUs. Example: „6204-2RS SKF“ or „Servo drive XYZ-1234 datasheet.“ Suitable format: Product, configurator, or inquiry page.
Local: The user is searching in their surroundings. Recognizable by location references and proximity keywords: „...near me,“ „...[city],“ „opening hours,“ „directions.“ Example: „hairdresser Munich“ or „rent construction equipment nearby.“ Expected results include Google Maps, opening hours, and correct location data. Suitable format: location or branch page with consistent NAP data (Name, Address, Phone) and a well-maintained Google Business Profile. This type corresponds to Google's own „Visit-in-Person“ category (see below) and is relevant in B2B for manufacturers with factory, service, or sales locations, less so for purely supraregional providers.
The first three types are based on academic research; commercial investigation and the separate Local category have supplemented SEO practice. The next section explains both threads.
What types of search intent lead to which decisions
Typology is not an end in itself: it determines which content format a page must support.
An informational request belongs on a glossary or how-to post, a navigational one on a brand or product overview page, a commercial investigation request on a comparison or solutions page, and a transactional one on a product, configurator, or inquiry page.
Anyone who places a data sheet where the SERP expects a comparison won't rank, regardless of optimization efforts.
So those looking for „food packaging“ usually don't immediately need a Online store, ..., who sells him machines. The user first needs information: What should I pay attention to? What options are there?
The Classic Three-Part Division by Broder and the Extensions of SEO Practice
Andrei Broder in 2002 „A Taxonomy of Web Search (SIGIR Forum) three categories were introduced: informational (wanting to know something), navigational (navigating to a specific page), and transactional (performing a transaction such as a purchase, download, or map retrieval).
In his log file analysis of 1,000 AltaVista queries at the time (Google wasn't always #1), the queries were broken down into approximately 20 % navigational, 48 % informational, and 30 % transactional queries.
Rose and Levinson refined the model in 2004 hierarchically, replacing the „transactional“ category with a broader „resource“ category; they identified approximately 62 % informational, 13 % navigational, and 24 % resource.
A subsequent large-scale log file analysis by Jansen et al. (2008, Penn State University) found over 80,000 informational queries and approximately 10,000 navigational and transactional queries each. The figures vary depending on the method and classification scheme—but one thing remains clear: informational queries account for the largest share.
A recent analysis by SE Ranking (2025) confirms this picture, albeit with shifted weightings: approximately 70 % of the keywords examined are informational, 22 are commercial, 7 are navigational, and 1 is transactional—the commercial share has grown significantly since the early studies, whereas transactional individual queries are rare, which aligns with the part number logic discussed below.
SEO practice has supplemented the academic model with two categories not found in the original studies: commercial investigation—the phase where someone compares providers, products, or solutions before buying or inquiring—and local, location-based search.
The narrower four-type model (informational, navigational, commercial, transactional) without the local category is still widespread; the five-type model used here cleanly separates local because location-based queries require their own content format and optimization signals. Neither of the two extensions has a unified scientific origin.
In the B2B industrial context, the majority of substantial inquiries fall into commercial investigation and informational categories. „Servo drive selection criteria“ is commercial, „Servo drive operating principle“ is informational, „Servo drive food industry manufacturers“ is commercial with a strong supplier research component, and „Servo drive XYZ-1234 datasheet“ is transactional in the narrower industrial sense (see part number searches below).
Google's Classification: Know, Do, Website, Visit-in-Person
Google itself classifies in the Search Quality Rater Guidelines four categories: Know (informational searches), Do (actions including purchasing), Website (direct brand or URL searches), and Visit-in-Person (directions, physical business locations).
In addition, Know-Queries are divided into Know and Know-Simple. The latter are factual short answers that mostly appear directly in the Knowledge Graph. Do-Queries include „Device Action“ queries such as voice commands.
At their core, both models lead to the same view: the SEO view classifies by content purpose, while the Google view classifies by expected user outcome. The categories largely overlap.
Know corresponds to informational, Do encompasses transactional and commercial, Website corresponds to navigational, and Visit-in-Person is the Google equivalent to the Local category from the SEO model.
The Google interpretation becomes practically relevant through the „Needs Met“ rating: according to Google's own definition, the usefulness of a result depends on how completely it fulfills the intention interpreted from the query.
It follows that the Rule of thumb: The SERP shows how Google has interpreted the search intent. So it’s best to just search for the keyword; that usually makes it clear right away what kind of information is being sought. Then just cater to that.
A page that formally has all the keywords but does not meet the interpreted intent will not rank.
The same principle is seen in user behavior as "pogo sticking": if someone clicks a result, immediately returns to the SERP, and selects the next one, it signals a gap between search intent and page content.
Whether Google directly uses this signal as a ranking factor is debated, but it's considered likely.
How to reliably determine the search intent of a keyword
Key takeaway: The most reliable source is the SERP (Search Engine Results Page) itself, not your own assessment.
Google has already formed an intent hypothesis for each keyword and made it visible in the top 10 results. Whoever predominantly sees glossary entries in the top ten organic results for a keyword will not rank with a product page, regardless of what the keyword suggests at first glance.
Three signals structure the SERP analysis:
- SERP Features as Intent Indicators Featured Snippets and People Also Ask indicate informational intent; Shopping boxes, product carousels, and pricingDisplay transactional; Knowledge Panels for navigational brand searches or know-simple queries. AI Overviews indicate that Google considers the query summarizable — this reduces the click-through rate for traditional results.
- Top content formats in the top 10: Do how-to guides, comparison tables, product pages, manufacturer data sheets, or forum discussions outweigh each other? The format pattern is the strongest individual signal.
- SERP overlap as a metric: The more identical URLs Google shows in the top 10 for two search queries, the more similar Google considers the underlying intent to be. In practice, this means: Two keywords with high URL overlap do not need two separate pages; they can — and generally should — be served on the same page. If the overlap is close to zero, the keywords, despite their semantic similarity, belong on separate pages.
The method breaks down at one point: For long-tail queries with very low search volume, Google sometimes provides SERPs that arise more from a lack of matching results than from intent interpretation. In an industrial context, this affects many highly specific technical queries—here, analyzing adjacent, higher-volume keywords is the more viable approach.
What to do when a term has multiple search intents
Search queries such as „beetle“ or „CNC machine“ yield mixed search intent in the SERP: The top 10 results include manufacturer product pages, Wikipedia/glossary entries, comparison sites, and application guides all at once. This is not a classification error, but rather Google’s acknowledgment that the term encompasses different user groups with different intentions.
The temptation to cover everything on one page regularly leads to content that lacks the right depth for any intention. A tiered content architecture along the buyer journey is more sustainable: a hub page that explains the term and visibly links sub-intentions (how it works, selection criteria, applications, products/product lines, inquiry).
Each sub-page then targets a narrower, clearer query: „frequency converter function“, „frequency converter sizing“, „frequency converter food industry“. This way, each term can rank cleanly, while the hub page covers the ambiguous term itself.
Which sub-pages are necessary is again decided by SERP overlap: If two narrower queries deliver largely the same URLs, one page is sufficient; if the SERPs diverge, they belong separately.
How generative search changes search intent
Generative search systems (like ChatGPT, Perplexity, Google AI Overviews, or Gemini) process queries differently than traditional SERPs. They break down a query into sub-questions (Query Fan Out), they draw on user context and predict the actual task behind the words, rather than primarily delivering keyword matches. In the SEO industry, a sixth intent type is spoken of, the so-called generative search intent: queries that directly aim for a summarizing, synthesizing answer instead of a list of sources.
In practice, two shifts are relevant. First: Traditional informational queries increasingly end in an AI response, without the user clicking on a source.
For glossary content, this means that the traffic-driven value decreases, while the citation value (being mentioned in the AI answer) increases. Second, remaining clicks from AI search are more intent-driven according to industry observations. This is because the user has already seen a recommendation and is specifically looking for confirmation, details, or providers.
Consequences for B2B prioritization: Informational top-of-funnel content will be geared towards quotability by LLMs—clear definitions, unambiguous sources, precise facts, structured statements. Commercial investigation and transactional content will retain their traditional SEO value because these are still clicked on: comparisons, configurators, specification pages, inquiry paths. Whether content targets KI citation or SERP clicks thus becomes an intent question in itself—no longer just relating to the user, but to the system that processes the request first.