I am currently obsessed with brand visibility in LLMs. Often called generative engine optimization (GEO) or AI optimization (AIO), thoughts about this topic are randomly hitting my brain.

Let me quickly explain.

I love Star Wars (nerd alert). Each holiday season I rewatch one movie – this year, it was The Empire Strikes Back. About an hour in, you hit that iconic scene where Yoda says, “Fear is the path to the Dark Side. Fear leads to anger, anger leads to hate, hate leads to suffering.”

No, GEO/AIO is not the dark side (unless you’re an SEO purist). What Yoda is talking about is inevitability. While watching that movie, I started thinking about the inevitability of LLM brand visibility. I couldn’t stop thinking about a stat I’m constantly parroting to teammates and clients, “By 2028, 75% of all search traffic will be through AI and LLMs.”

As teams roll into 2026, I’m having weekly discussions about GEO/AIO. All understand the inevitable impact LLMs will and are having on search, but many are asking how to get started. So, I wanted to share my approach.

A quick primer.

To ensure we’re speaking the same language: GEO/AIO is the practice of improving how a brand or product shows up in AI-driven query outputs. The goal is to ensure a brand appears accurately, favorably and consistently when users ask AI-powered systems questions relevant to your space.

As LLMs increasingly become the cornerstone for search, brands must understand how they perform and put a strategy in place to grow and maintain their visibility. A big part of that strategy involves demonstrating knowledge, expertise, authoritativeness and trustworthiness. The sweet spot is to collaborate with both your SEO and PR team. SEO for backend and some content optimization; PR for press releases, thought leadership content, earned media placements, social media optimization and more.

Do or do not. There is no try.

Getting down to brass tacks, I want to share my general approach for diving in and evaluating a brand’s LLM visibility.

1. You need data – full stop. If a brand already has a GEO/AIO tool, great. If not, find an agency that can help. (For example, Voxus uses Semrush Enterprise AIO.)

2. Start by evaluating three models to get a baseline. I recommend Chat GPT, Gemini and Google AI Overview. You can also look at others like Claude, Perplexity or Grok.

3. Identify your top 10 non-branded keywords/phrases. Those are then used by your tool to generate prompts that are measured against the models. I recommend starting with 40-50 prompts (you can expand later). Focus on brand/market initially, expanding into niche topics later.

4. Let the tool collect data for a couple of days then look at the high-level data first. These tools will show you the overall share of voice (which is often a calculation of mentions, frequency and position), your source visibility, referral traffic from LLMs, an overall visibility score, etc. This will give you an initial benchmark of performance.

5. Go a layer deeper. Look at each model and the prompts to see if your brand was used either as a mention (your brand’s name or product), linked source (your URLs), or if your source material was used for a response, but your name or a link weren’t included. You can also get a trove of other information such as top sources, sentiment, competitor information, key concepts, back-end schema errors, content audits, etc. – all vitally important for the next step.

6. Once you understand the state of your LLM performance/data, you can begin to create a strategy for improvement. For example, backend website schema errors might need to be resolved so LLMs can scrape your site properly. Source information will show you where to prioritize efforts or build micro programs (e.g., LinkedIn, Reddit, YouTube, Wikipedia, earned media sites, etc.). All those data elements become powerful signals that will inform your plan.

7. Once you formulate your strategy and tactics, execute for 3-6 months (depending on the goal). As mentioned before, data is dynamic, so you want to look at results in aggregate.

It’s inevitable that brands will have to optimize for LLM visibility. People are already asking AI systems which vendors to trust, which tools to evaluate, and which companies understand their challenges. If your brand is missing from those answers, someone else fills the gap.

The good news is that this is measurable. It is improvable. And it rewards discipline.

Brands that invest early in understanding how they appear across models will make better decisions about content, PR, and authority building. They will know what signals matter and where to focus…and they’ll be able to track progress over time.

Need help, reach out to us.

Side Quest

In case you didn’t believe my obsession with Star Wars and GEO, yes, I checked Yoda’s LLM brand visibility.

I ran a report using StarWars.com as the brand and Yoda as the concept (or product if you will). Confirming how ridiculous this exercise was, I can share with you (drumroll please), that StarWars.com has pretty decent brand visibility across most models (not a surprise). However, to my disappointment, folks are not spending all day asking these models about Yoda (his species, planet, object-subject-verb order when speaking, etc.). You’re welcome.