How can we identify user intentions specific to AI?

What People Ask an Assistant That They Don't Type Into Google, the Six Types of Intent, Where to Find Them, and How to Use Them to Create Content

Key Takeaways

  • The difference: In Google, you type in three words; with a virtual assistant, you describe a situation; the intent is explicit, contextualized, and often decision-driven.
  • Six types: understand, compare, choose, perform (a procedure), verify, delegate (writing, calculating, planning).
  • Where to find them: customer questions, Google's "other questions asked," suggestions from assistants, Reddit and forums, GEO tracking tools, and your own conversations with AI.
  • Method: A list of 30 to 50 actual questions, organized by type and stage of the program, linked to a page.

A user who types “heat pump price” into Google and one who asks ChatGPT, “I live in a 120-square-meter house in Loire-Atlantique that’s heated with fuel oil—is a heat pump cost-effective, and how much does it cost?” ” have the same underlying intent, but the second one expresses it fully. AI assistants are receiving longer, more contextualized requests that are closer to the decision-making process. Identifying these intents is the starting point for the GEO : They decide what content to produce and how to write it. This article describes how they differ, the six types, the sources, and the method. It complements Keyword Research for GEO.

How do AI intents differ from Google search queries?

Google SearchAsk an assistant
Length2 to 5 words15 to 60 words, sometimes several rounds of conversation
BackgroundImplicit (the user then refines the search)Specify: situation, constraints, budget, location
FormKeywordsQuestion or command in natural language
WaitingA list of pages to browseA concise response, often a recommendation
Stage of the routeAllAbove all, understand, compare, and choose; the transaction takes place elsewhere
What the response citesTen Links3 to 10 sources, one sentence each

Implications for content: A page that responds to a three-word query with a generic article will not provide an answer to a forty-word question; it is necessary to cover the situation, its variations, and the decision criteria. The search engines are described in Understanding AI Engines.

The Six Types of Intentions in Assistants

TypeExampleRelevant contentStep
Understanding"Explain the difference between SEO and GEO to me."Definition, Guide, Comparison ChartDiscovery
Compare"Semrush or Ahrefs for a 5-person agency?"Comparison with Criteria, Prices, and Use CasesConsideration
Choose"Which SEO provider in Nantes would you recommend for an SME?"Company page, reviews, listings and media mentions, Google business profileDecision
Do"How do I set up key events in GA4, step by step?"Numbered procedure, catches, trapsUsage
Check"Is it true that Google doesn't penalize AI-generated content?"Fact-based, sourced response; nuancesAll
Delegate"Write an SEO brief for my services page," "calculate my ROAS"Models, calculators, and methods that AI draws upon and citesUsage

The “choose” and “compare” types have the highest commercial value: this is where assistants mention brands. The “understand” and “verify” types build the authority that leads to being cited elsewhere.

Where to Find the Real Intentions

  1. Your customers: Questions asked before making a purchase (sales representatives, customer service, forms, phone calls). They are phrased as if you were speaking to an assistant.
  2. The assistants themselves: Ask your initial questions to ChatGPT, Perplexity, and Gemini; note the rephrasings, suggested follow-up questions, and related topics they provide.
  3. Google: "Other questions asked," suggestions, and queries phrased as questions in Search Console (regex filter for "how," "why," "which").
  4. Communities: Reddit, industry forums, LinkedIn groups, and Facebook: questions there are written in the users' own words, and AI systems read these sources.
  5. GEO monitoring tools: Some provide lists of questions by sector and track who is mentioned; see track your visibility on ChatGPT and Perplexity.
  6. Your own conversations: The requests you make to AI systems in your line of work reveal the intentions of your peers.

The Method

  • Collect 50 to 100 actual questions, retaining the original wording.
  • File by type (the six listed above) and by stage of the journey.
  • Group Questions that call for the same answer: one page per group, not per question.
  • Prioritize : Start with the "choose" and "compare" groups related to your offering, then move on to "understand" for topics where you have expertise.
  • Link Each group should choose an existing page to revise or a new page to create, with the main question as the subtitle and the direct answer below it; see Conversational Optimization.
  • Test Every month: ask the assistants questions, write down their answers, and make adjustments.

Common Mistakes

  • Translate the Google keyword plan as is: the phrasing and context are missing.
  • Don't limit yourself to "understanding": the definitions cite examples, not clients; you also need to "compare" and "choose."
  • Come up with questions: those from customers and communities are always more specific than those brainstormed in meetings.
  • One page per question: Tutors prefer a single page that covers the situation and its variations.
Tip: Ask three recent customers what they typed into ChatGPT or Google before contacting you—word for word. These three sentences are more valuable than any tool for understanding the purpose your web pages should serve.

How GreenRed Can Help You

Instead of juggling multiple tools, GreenRed's GEO and AI module brings these metrics together in a single dashboard, compares them over time, and shows you which actions to prioritize. You can try it for free—no credit card required—on the page Rates.

Frequently Asked Questions

Do AI users have different intentions than Google users?

The underlying intents are the same (understand, compare, choose, act), but the way they are expressed changes: long, contextualized requests in natural language, with an expectation of a concise response or recommendation. The assistants primarily focus on intents related to understanding, comparing, and choosing.

How can I find out what people are asking ChatGPT about my industry?

There is no “keyword planner” for ChatGPT. Sources are indirect: your customers’ questions, suggestions and rephrasing from assistants, questions in Search Console and Google, communities (Reddit, forums), and GEO tracking tools that provide lists of questions by industry.

Should we create a page for each question?

No. Group together questions that call for the same answer and cover the group on a single page, with the main question as the subheading and a direct answer below it, followed by the variations. Assistants cite the pages that address a situation as a whole.

Which intentions have the most commercial value?

The “compare” and “choose” intent categories: this is where virtual assistants mention brands, tools, and service providers. These are influenced by honest comparisons, clear company pages, reviews, and mentions in the sources that AI systems read. “Understand” search intents build the authority needed to be cited.

Track your visibility on AI platforms

GreenRed monitors whether and how ChatGPT, Perplexity, and Gemini mention your brand and your competitors.

Try GreenRed for free

Related Articles