Evaluating & Optimizing Brand Visibility in LLMs

The idea is this:

Since LLMs are becoming a go-to source for research, we need processes to evaluate how our brands are appearing. That means checking not just if your brand shows up, but how it's framed, what it's associated with, and whether competitors are owning more of the narrative.

Here’s the rough process I’m testing to evaluate and optimize brand visibility in LLMs (using Patagonia as an example):

Step 1 - Build Personas:

Create realistic customer personas based on key audiences. Include who they are, what they care about, what they’re searching for. Building personas is important because they help you test LLM responses through the lens of real-world user intent.

Here are what some personas could look like for Patagonia:

• Olivia: 32, avid hiker, cares about environmental impact
• Tim: 41, adventurous dad, wants to spend time with his kids
• Jake: 26, casual explorer, into practical and fashionable clothing

Step 2 - Test the Competitive Landscape:

Start with broader, brand-neutral prompts your personas might ask when first researching options. For example:

• Olivia: “What are the most sustainable outdoor clothing brands?”
• Tim: What are the best jackets for cold-weather camping with kids?”
• Jake: “Which outdoor brands have hiking clothes that are functional but also look good?”

See which brands show up most often, how they’re positioned, and what themes they’re associated with (e.g. sustainability, durability, style).

Step 3 - Test Your Brand in LLMs:

Use those personas to ask questions in LLMs and see how your brand shows up (if it does). Look at tone, accuracy, and overall positioning.

• Olivia: “How eco-friendly is Patagonia?”
• Tim:“Is Patagonia good quality for family camping trips?”
• Jake: “Is Patagonia a stylish brand or just practical?”

Step 4 - Research What People Are Asking on the Web:

Step 5 - Build a List of Key Entities:

Start tracking which concepts and entities matter in your space, (product types, use cases, pain points, etc.) so you can later check if your content reflects them.

Examples for Patagonia might include:

• Product types: base layers, insulated jackets, hiking pants
• Values: sustainability, climate activism, durability
• Features: recycled materials, lifetime guarantee, repair program
• Related terms: technical gear, outdoor lifestyle, fair trade

Step 6 - Create or Update Content Accordingly:

Instead of just focusing on high search volume keywords, use your insights to strengthen how your brand shows up semantically. If LLMs associate key topics with competitors or miss your brand entirely, update or create content that builds those connections. Focus on questions your audience is asking, reinforce themes uncovered in your testing, and make sure your content reflects the semantic relationships that matter (like pairing your brand with important product types, values, and use cases).

Step 7 - Run Entity Salience Tests:

Entity salience is a measure of how important or central a specific entity (such as a person, brand, product, place, or concept) is within a piece of content.

Use tools like Google Cloud NLP to measure the entity salience in your new content (If you're curious on how to do that, read this guide.)

This approach focuses on strengthening the semantic relationships that LLMs are more likely to pick up on. Since LLMs learn from patterns across web content, helping them understand how your brandrelatesto certain topics could improve how you show up in responses.

Step 8 - Keep Checking!

To figure out if your content updates are working, you need to keep checking. Regularly re-test LLM outputs using your personas and prompts. Track improvements in mention frequency, tone, and topical alignment for both your brand and key competitors.

A good way to do this is to use Ahref's newly released Brand Radar, or SEMrush's Position Tracking tool!

These can help you:

• See how often your brand appears in tools like ChatGPT, Perplexity, or Gemini.
• Track shifts in rankings or mentions over time.
• Benchmark visibility against competitors.

These tools complement manual testing by giving you a broader, more automated view of your brand’s presence across generative AI platforms.