Do you have a competitive GEO problem? How to know if AI recommends you or your rivals

"Why is AI recommending our competitor over us?" If that question has landed on your desk, you are not alone. Marketing leaders everywhere are being asked whether AI tools mention their brand more or less often than the competition. The honest answer is harder to pin down than most people expect.

Here's the catch: the AI visibility tools promising to rank you against rivals often deliver conflicting, inconsistent results. I recently attended a webinar hosted by Trust Insights and it covered everything from why AI rankings are unreliable, what actually drives whether AI recommends you and which signals you can measure with confidence. Let’s dive into what each one means.

Why AI visibility rankings can't be trusted

Generative engine optimization, or GEO, is the practice of making your brand visible and recommendable inside AI tools like ChatGPT, Gemini and Google's AI features. It sounds a lot like SEO, and there's plenty of overlap. But one thing makes GEO far messier: AI systems are probabilistic.

That means the same prompt can produce different answers from one moment to the next. Ask an AI tool which brand it recommends today, and you may get a different result tomorrow or even on the next attempt.

Personalization makes it worse. Google, for example, pulls everything it knows about the person asking. Search history, location, past behavior, all of it shapes the response. So, when a vendor hands you a tidy ranking that says you sit third behind two competitors, you should be skeptical. That number reflects one probabilistic snapshot, filtered through one set of personalization data. It is not a stable scoreboard.

This is why so many AI visibility tools disagree with each other. They are each sampling a moving, personalized target and reporting their version as fact.

The smarter question isn't "What's our ranking?" It's "Do we actually have a GEO problem, or does the whole industry have one?" Sometimes the issue is yours to fix. Sometimes the category is simply invisible to AI, and that's a very different conversation with your stakeholders.

The three phases of GEO

Instead of chasing rankings, break the problem into three phases. Each one answers a distinct question and each points to a different fix.

• Phase 1: Presence -- What does the AI model already know about your brand, before it searches anything? Large language models are trained on enormous amounts of text. Presence is about what the model absorbed during that training. When someone asks about your category and the model answers from memory, does your brand come up?  If you barely register here, it usually means your brand wasn't well represented in the content the model learned from. Building presence is a long game. It comes from consistent mentions across the web, recognized associations with your topic and a clear, repeated identity.

• Phase 2: Awareness -- When an AI tool runs a live search, can it find your content? Many AI tools now search the web in real time before answering. Awareness measures whether your content surfaces during that live retrieval. This phase sits much closer to traditional SEO. If your pages aren't crawlable, indexed or relevant to the query, the AI simply won't find them. Here’s the good news: awareness is more fixable than presence. It rewards the same fundamentals that have always supported discoverability, including clean technical setup and content that matches real questions people ask.

• Phase 3: Relevance -- Once the AI finds your content, is it structured so the model can actually use it? Being found isn't enough. The AI has to be able to read, parse and pull meaning from your content. Relevance is about structure: clear headings, direct answers, well-organized information and content that's easy for a machine to interpret and quote. Think of it this way. Presence gets you into the model's memory. Awareness gets you found in the moment. Relevance gets you used in the answer. You need all three.

Measurable signals you can actually trust

If rankings are unreliable, what should you measure instead? Focus on signals that are stable, observable and repeatable. Here are the practical approaches covered in the webinar:

Crawl your site for GEO issues: Use a tool like Cloudflare Radar to crawl your website and surface technical problems that hurt AI discoverability. This directly supports the awareness phase. If AI crawlers can't access or interpret your pages, nothing else matters. A crawl gives you concrete issues to fix rather than a fuzzy score to argue about.

Watch named entities in your inbox: Your email inbox is a surprisingly useful diagnostic tool, and Gmail tested as the clearest signal. The technique relies on named entity recognition, meaning the brands, people and topics that AI systems identify and connect.

Here's how to use it:

• Search your inbox for your brand and your competitors as named entities.

• Notice which names appear, how often and in what context.

• Compare your brand's footprint against rivals.

This gives you a window into how often your brand shows up in the kind of content that shapes AI understanding.

• Subscribe to the right newsletters and Substacks: Sign up for the newsletters and Substacks that matter in your industry. Better yet, ask an AI tool which publications it would recommend subscribing to in your space, then subscribe to those. Why does this help? Those publications are influential sources. The brands they mention build named-entity strength over time, which feeds both presence and your competitive standing. Reading them also tells you which competitors are earning that visibility while you watch your inbox signals.

• Run an AI visibility check: Enter your website into an AI visibility checker that reviews the full set of criteria behind AI visibility. Trust Insights offers one at trustinsights.ai/aiview. A tool like this looks across the factors that matter rather than handing you a single shaky ranking, which keeps your diagnosis grounded in evidence. The biggest mistake leaders make is reducing GEO to one rank. A number feels reassuring. It fits neatly in a slide. But it hides the probabilistic, personalized reality underneath and it sends teams chasing a target that won't sit still. The fix? Report on phases and signals instead. Tell stakeholders where the gap is (presence, awareness or relevance) and which measurable signals you're tracking. That story is more honest, more useful and far easier to act on.

GEO matters because AI tools increasingly shape which brands people consider, and your leaders already feel the pressure to compete there. The trap is trusting rankings that shift with every prompt and every user. Here's what to do instead. First, stop reacting to single rankings and start diagnosing by phase. Ask whether the AI knows you (presence), can find you (awareness) and can use your content (relevance). Second, lean on signals you can actually measure: crawl your site for technical GEO issues, study named entities in your inbox and subscribe to the publications that influence your category.

Your next step is simple. Pick one phase where you suspect a gap, run a crawl or an AI visibility check to confirm it, and bring your stakeholders a clear, evidence-based answer instead of a number nobody can trust.

Let AOE help you turn a vague "AI doesn't recommend us" worry into a clear, measurable plan. Reach out to us today at aoeteam.com