Most of what gets written about Generative Engine Optimization right now focuses on the LLMs — how ChatGPT cites sources, what Google’s AI Overviews favor, which platforms weight Reddit versus G2. That’s all useful, but it skips the most important variable in the entire discipline.
The variable is the buyer.
Different buyers use AI tools completely differently. A dentist researching a new sterilization system prompts ChatGPT with different language, validates the answer through different sources, and trusts different signals than a CFO evaluating treasury management software. A software developer asking Claude about an API library prompts with technical specificity that an executive marketing leader asking about agency partners never will.
The signals that get a brand cited inside an AI answer aren’t universal. They’re specific to how a particular buyer is using a particular tool to evaluate a particular kind of decision. Any GEO strategy that doesn’t start with mapping the buyer’s actual behavior is optimizing for an average buyer who doesn’t exist.
This piece is about the methodology we use when we start GEO work with a client. Every consulting engagement we run starts here, before any audit, before any tactical recommendations. If you take one thing from this piece, take this: the way your buyer actually uses AI is the most important input to your GEO strategy, and most teams skip it because they assume it’s similar to how they themselves use AI. It usually isn’t.
The four dimensions that change everything
When we map a buyer’s AI usage, we look at four things. Each one shifts the GEO strategy substantially.
1. Which tools they actually use
Most marketers default to thinking about ChatGPT because that’s what they use. But the AI tool distribution among buyers is more variable than the marketing trade press makes it sound.
CFOs and finance leaders are increasingly using Claude. Software engineers prefer Claude or specialized tools like Cursor for technical decisions. Marketing operations people use ChatGPT and Perplexity roughly equally. Sales leaders often use whatever their company’s enterprise license is, which is increasingly Microsoft Copilot. Younger buyers use TikTok and Reddit’s native search alongside or instead of traditional AI tools.
This matters because the citation behaviors of these tools differ significantly. Claude doesn’t surface citations in its base interface; ChatGPT does (sometimes); Perplexity always does and prominently; Google’s AI Overviews show source links but with different weighting. A GEO strategy optimized for ChatGPT might be invisible to Claude users — and the implications for your business depend entirely on which buyers in your category use which tool.
The exercise we run with every client: have your team interview five customers about which AI tools they used during their evaluation process. Don’t ask “do you use AI?” Ask “the last time you researched a vendor in our category, which AI tools did you open and what did you ask them?” The answers usually surprise the marketing team.
2. How they prompt
The way a buyer constructs a prompt determines which sources the LLM pulls. A buyer who types “best marketing automation platforms” is asking a broad-category question, and the LLM will draw from listicles, analyst reports, and category-level reviews. A buyer who types “Marketo vs HubSpot for mid-market B2B SaaS with a 12-person marketing team” is asking a specific-comparison question, and the LLM will pull from very different sources — peer-review sites, Reddit threads, specific vendor comparison content.
We see four broad prompt patterns in our consulting work, each pulling different source mixes:
Discovery prompts — broad-category questions. “Best X for Y.” Sources skew toward listicles, third-party rankings, analyst content. To show up here, you need to be in the listicles.
Comparison prompts — specific vendor-vs-vendor questions. Sources skew toward review sites, Reddit threads, side-by-side comparison content. To show up here, you need review-platform presence and user-generated content.
Validation prompts — buyers who have a shortlist and are checking it. “Is [Vendor X] good for [specific use case]?” Sources skew toward case studies, customer testimonials, deep-dive reviews. To show up well here, your customer marketing has to feed the AI tools.
Diagnostic prompts — buyers who don’t know what they need yet and are asking the AI to help them figure it out. “I’m a [role] at a [company size], what should I be looking for in a [category] solution?” Sources skew toward thought leadership, educational content, frameworks. To show up here, you need to be the voice teaching the buyer how to think.
Most companies optimize for one of these and ignore the others. Most buyers use all four during a single buying cycle.
3. What context they bring
This is the dimension nobody talks about. ChatGPT, Claude, and Perplexity all weight prior conversation context. They also weight information the user has previously shared about themselves, their company, their role, their industry, their priorities.
We had a services-provider client who was being routinely recommended against by ChatGPT when US-based prospects asked for vendors in their category. The reason: their website emphasized their global services capability so heavily that ChatGPT had encoded them as “global, not US-focused” and was specifically excluding them from US-prospect recommendations. The client did business primarily with US companies. The brand positioning was working against the buying journey.
The fix was a website update making the US focus more prominent, plus a piece of first-party research that gave LLMs corroborating signal that the client served US customers. Within six weeks, citation patterns shifted.
The exercise: have your team prompt ChatGPT (or your buyer’s preferred tool) with questions about your category, once with no prior context, once with a fake but plausible buyer context (“I’m a CMO at a mid-market B2B SaaS company in the US looking for X”). Compare what the LLM cites in each case. The difference tells you what context the tool is weighting, and where your brand positioning might be working against you.
4. What signals they actually trust
Different buyers validate AI recommendations through different sources. Technical buyers Google a recommendation and look at GitHub, documentation, and technical Reddit threads. Marketing buyers look at G2 reviews, LinkedIn posts from peers, and the vendor’s own website. Executive buyers look at analyst reports and named-customer case studies. Younger buyers look at video — YouTube reviews, TikTok mentions, creator content.
Whatever signals your buyer trusts is where you need GEO presence beyond getting cited in the LLM in the first place. A citation in ChatGPT that the buyer can’t validate through their preferred trust signal is a citation that doesn’t convert.
What this means for how to actually do GEO
If you’re a marketing leader trying to figure out where to invest in GEO right now, the practical implication is this: don’t start with the platforms. Start with the buyer.
Specifically:
- Interview 5-10 real customers about how they used AI during their last vendor evaluation. Which tools, what they prompted, what they validated through.
- Map the four prompt types for your category. Discovery, comparison, validation, diagnostic. Know which ones your buyers actually use and in what proportion. Most companies are surprised by what they find.
- Test your own brand in the buyer’s preferred tools using the prompts your buyers actually use. Don’t generalize from your own ChatGPT usage. Test what your buyer does.
- Map context dependencies. What does the LLM assume about a buyer in your category before they ever prompt? What is your website teaching the LLM that might be working against you?
- Identify the trust signals your buyer uses to validate AI recommendations, and make sure you have presence there.
Then, and only then, start thinking about platform-level GEO tactics.
A lot of GEO writing skips straight to “publish more Reddit content” or “optimize your YouTube descriptions.” Those things might be right for your buyer or they might not be. The work that makes them right is the buyer-mapping work.
What we know about how this changes by category
A few patterns we see consistently across our consulting work:
B2B SaaS: Marketing leaders heavily use ChatGPT and Perplexity. CFOs increasingly use Claude. Comparison prompts are the highest-leverage prompt type. Review site presence (G2, TrustRadius) is critical.
Professional services (legal, accounting, consulting): Executive buyers use Claude and ChatGPT roughly equally. Diagnostic prompts dominate — buyers often don’t know what they need and use AI to figure it out. Thought leadership content has outsized impact here.
Industrial / manufacturing: Buyers skew toward Google AI Overviews more than chat-style AI. Discovery and validation prompts dominate. Analyst and industry-publication presence matters disproportionately.
Healthcare / medical (the dentist example): Buyers tend to trust AI recommendations less and validate heavily through professional networks. ChatGPT and Perplexity are the dominant tools but the validation step matters more than the recommendation step. Community presence in industry-specific forums is unusually important.
These patterns are starting points, not universal truths. The point of doing your own buyer mapping is that your industry’s patterns might match these and might not. The cost of generalizing is optimizing for a buyer who doesn’t exist.
Frequently Asked Questions
What is the difference between SEO and GEO?
SEO (Search Engine Optimization) focuses on ranking in traditional search engine results pages. GEO (Generative Engine Optimization) focuses on increasing the likelihood that your brand is cited and recommended within AI-generated responses from tools like ChatGPT, Claude, and Google AI Overviews.
How do I get my company cited by ChatGPT?
AI models cite sources they deem authoritative and relevant to the user’s prompt. To be cited, your brand needs a strong presence on high-authority third-party sites (like G2, Reddit, and industry publications) and clear, structured data on your own website.
Which AI tools are B2B buyers using most?
While ChatGPT has the highest volume, professional buyers are increasingly using Perplexity for research-heavy tasks, Claude for analysis, and Microsoft Copilot for integrated workplace searches.
If you want to skip the buyer-mapping exercise and just see how your brand currently appears across the AI tools your buyers actually use, GNW Consulting offers a free competitive GEO audit. The audit includes buyer-prompt mapping for your specific category — we test the prompts your buyers actually use, not generic ones, and show you which platforms are citing your brand and which are citing your competitors instead.
Related reading: What Generative Engine Optimization Actually Is — the canonical definition piece. What we actually know about LLM citation signals — and what we don’t
