AI models like ChatGPT, Claude, and Gemini are reshaping how people find and evaluate information. Instead of searching Google and clicking through results, we now ask questions directly to a language model - and get a condensed answer, often without visiting a single website. This shift has given rise to a new discipline: LLM optimization. Like SEO, it’s about visibility - but in a new ecosystem where sources and data accessibility determine which voices the models “trust.”

Why LLM optimization matters and why it’s time to act

LLMs (Large Language Models) draw on massive amounts of text to form their understanding of the world. When a user asks, “Which Danish agencies are good at digital strategy?”, the model selects its answers from the sources it knows - public websites, articles, case studies, interviews, and increasingly, third-party platforms like media outlets, industry blogs, GitHub, LinkedIn articles, and dataset repositories. If your company isn’t represented in those sources, you’ll be invisible to AI - no matter how strong your SEO is.

Right now, there’s a short window of opportunity because few companies are working systematically with LLM optimization. Getting included early in training datasets and citation chains gives you an advantage that’s hard to catch up to later.

How to start working with LLM optimization

Here’s a step-by-step guide to help you get started - or how we can help you get there.

1. Make a list of 10–20 key prompts

Start by thinking like your potential clients or partners. What questions might they ask an AI where you’d want your name to appear in the answer?

Examples:

• “Which digital agencies in Denmark work with AI?”

• “How do you create a digital strategy for B2B?”

• “What are the best cases in online brand activation?”

• Create a list of 10–20 of the most relevant prompts. That’s your LLM keyword universe.
  
  

2. Research and collect sources

Enter each prompt into ChatGPT (or another model) and note which sources it mentions or cites. Write down your observations so you can identify patterns.

For each prompt, consider noting:

• Which sources are mentioned?

• Any notes - for example, if you’re already included, or if there’s something you could build on.

Example: When you ask, “Which agencies work with AI?”, the model might mention Bureaubiz.dk, LinkedIn articles, and Baernholdt.com. This shows that you’re already visible in some places - but that there are also third-party sources where it might be worth strengthening your presence.

By repeating this exercise across your key prompts, you’ll get a clear picture of where you’re visible today and where you should focus your efforts to appear more prominently in AI-generated results.

3. Map your competitors

Make a list of your main competitors and see whether they appear in the answers and why.

• Are they cited from their own case studies?

• Do they publish on third-party platforms?

• Are they mentioned in industry articles or reports?

This gives you a clear overview of their LLM footprint.

4. Categorize your sources

Once you’ve gathered your results, divide the sources into three categories:

• Third-party intermediaries (e.g., media outlets, blogs, directories, reports)

• Your own site

• Competitors’ sites

This distinction is important because your strategy depends on where you can most effectively influence visibility.

5. Reach out and build visibility

For third-party intermediaries:

• Reach out to editors and pitch case studies or expert commentary.

• Contribute articles or insights that add real value.

• Make sure your brand is mentioned with context, so LLMs understand your relevance.

For your own site:

• Create content that answers the key prompts directly.

• Use clear professional signals; explanations, examples, case studies, quotes.

• Use precise, natural language; LLMs prioritize clarity and comprehension over keyword density.

For competitors’ sites:

• You’re unlikely to get competitors to mention you on their sites, but it’s still worth studying how they structure their content and what might make them appear in model-generated answers. You can then consider whether a similar strategic approach could be relevant for you.

• Never copy content - it’s poor form, it’s plagiarism, and it won’t help you rank in model outputs.
  
  

6. Evaluate and repeat

Repeat the process every 3–6 months. New models and updated training datasets constantly reshape the landscape — much like SEO once did. By monitoring your progress over time, you can measure whether your brand is becoming more visible in AI models.

For most companies, LLM optimization is still experimental. But just as SEO became an established discipline, LLM optimization will become a natural part of every digital strategy.

If you want to be among the first to take control of your brand in the AI era, we’re here to help you get started.