Understanding search intent
Informational, navigational, transactional, or comparative: identifying the intent behind a query determines which content format actually stands a chance of being seen, read, and cited.
Search intent is the goal a person is pursuing when they type a query into a search engine or a prompt into an AI assistant like ChatGPT. Guessing it correctly determines the format, tone, and substance of the content — or the answer — that will actually satisfy that person.
The four main intent categories
Most queries fall into one of four categories.
Informational. The person wants to understand a topic, learn a concept, resolve a doubt. “What is GEO”, “how does a generative search engine work”, “why is my traffic dropping” are informational queries. This is the broadest category, and the one where long, educational content performs best.
Navigational. The person already knows where they want to go and uses the engine as a shortcut. “Google Search Console login”, “Semrush pricing”: there's little value in creating content here, unless you are the brand being searched for.
Transactional. The person is ready to act: buy, sign up, download. “Free trial SEO tracking tool”, “buy a .com domain name”. The content should remove the last barriers to a decision, not re-explain the basics.
Comparative. An intermediate intent, between informational and transactional: the person is weighing several options and looking for arguments to decide. “Semrush vs Ahrefs”, “best CMS for a professional blog”. These queries are particularly likely to be cited by generative engines, which tend to produce comparison tables.
Why this typology matters as much for GEO as for SEO
In classic SEO, misreading intent shows up as poor rankings: Google favors pages that match the expected result type for a given query. If the top ten results are guides and you publish a product page, you won't rank, no matter how good your content is.
In GEO, the stakes are different but just as structural. A generative engine doesn't rank ten links: it builds a single answer, usually drawing on three to eight sources. If it interprets the query as informational and your page reads like a sales brochure, it simply won't be cited — not because it's bad, but because it doesn't answer the right question. This is also why a mismatch in intent costs more in GEO than in classic SEO: there is no “page eleven” in a generated answer.
How to guess the intent behind a query
Three methods complement each other.
Look at the results already ranking. Type the query into Google and see what shows up at the top. Blog posts? The intent is probably informational. Product pages or e-commerce category pages? Transactional. Comparison tables? Comparative. The engine has already done part of the classification work for you, based on the behavior of millions of users.
Analyze the wording. Question words (“how”, “why”, “what is”) almost always signal informational intent. Action verbs (“buy”, “download”, “sign up”, “try”) signal transactional intent. Two brand names, or words like “vs”, “or”, “best”, signal comparative intent. A single brand name with nothing else signals navigational intent.
Ask a generative engine directly. Put the query to ChatGPT or Claude and observe the type of answer it produces spontaneously: a structured explanation, a comparison table, a step-by-step list. This is a concrete indicator of how the engine itself has categorized the request — useful for GEO, since that engine is exactly the one you're trying to satisfy.
Combined on a real example, these three methods reinforce each other: for the query “CRM for small business”, the Google results mix guides and comparison pieces, the phrasing contains no action verb but two generic terms, and a generative engine typically answers with a list of criteria followed by software examples. The diagnosis becomes clear: a mainly comparative intent, with an informational undertone.
The case of single-word, ambiguous queries
A query like “GEO” or “SEO”, reduced to a single generic word, is particularly hard to classify: it can just as easily reflect an informational intent (“what is this”) as a navigational one toward a specialized site. In that case, it's better to rely on the context of the site publishing the content rather than the query alone: a site already positioned around teaching the subject has every interest in treating this query as informational, then guiding the reader toward more precise topics once the definition is in place.
A single query can carry more than one intent
Many queries are mixed. “Best invoicing software for freelancers” is both comparative (several options at play) and transactional (the person is close to choosing). In that case, the best answer often combines a comparison table near the top with clear action links for each option, rather than forcing a choice between the two formats.
This is also why a generic query like “GEO” deserves to be broken down into sub-topics handled separately rather than a single catch-all article: this logic of splitting a topic up is at the heart of the Hub & Spoke architecture, which organizes content around a broad intent split into more precise ones.
Matching content to the intent you detect
Once the intent is identified, the content's structure follows almost mechanically:
- Informational intent → long, educational content, with clear definitions and concrete examples. This is the territory of the pillar page.
- Comparative intent → tables, objective criteria, the pros and limits of each option stated without excessive bias.
- Transactional intent → short content, focused on the decision, with social proof and a clear call to action.
- Navigational intent → little editorial effort needed; the real work is technical (a fast, well-indexed page).
Before even searching for keywords with tools like Google Search Console or free tools, mentally sorting each candidate query into these four families keeps you from producing content that will never find its audience — human or generative.