Using Entity Salience to Win in SEO and LLMs
Learn the role of entity salience in SEO, why it's important for building topical authority beyond keyword targeting, and how to use Google's NLP API.
I'd like to start off by saying that this isn’t a call to throw out keyword research (it's still very important). But if you're still optimizing content by simply picking high-volume keywords and placing them throughout the page, you're falling behind.
Google and large language models (LLMs) don't recommend or use content just because it contains the right terms anymore. They prioritize content that demonstrates actual understanding of the topic.
One of the clearest signals of that understanding is entity salience.
What is entity salience?
Entity salience refers to how important a concept is within a piece of content. An "entity" could be a product, brand, person, ingredient, or idea. If a term is mentioned frequently, used in context, and surrounded by relevant information, it's considered more salient. These terms could have search volume, but that isn't always the case.
Search engines and LLMs rely on this to evaluate whether your content aligns with what users mean when they search or ask a question.
Google’s Natural Language API provides a way to measure salience by assigning scores to entities in your content, but the concept is used across nearly every major AI model.
Real-world example: Nike
Say you're writing an article targeting the keyword "best nike running shoes".
A basic approach might include phrases like“best nike running shoes 2025”, “comfortable nike running shoes” or “nike shoes for long-distance running”.But when you analyze top-ranking pages, you might see additional highly salient entities like"carbon fiber plate","heel drop", or"midsole foam".
These are not just surface-level terms. They’re concepts that show expertise. If your content is missing these, it may not be perceived as authoritative or useful.
Why does this matter for LLMs?
LLMs don’t search for keywords. They interpret meaning based on patterns, relationships, and context. When your content includes the right entities and uses them in meaningful ways, it helps LLMs understand what your page is about and when it should be referenced.
If your content doesn't naturally incorporate the correct entities, AI models may misinterpret the topic of your content, or your brand or content may be skipped in responses. Writing with entity salience in mind increases your chances of being surfaced, cited, or summarized by AI tools built on LLMs.
For brands, that means greater visibility, stronger associations between your brand and key topics, plus more opportunities to shape how your brand is positioned.
How to use Google’s NLP API to analyze entity salience:
1. Head to Google Cloud Natural Language and scroll down to the "Try the API" section.
2. Paste your content
Take content from a top-ranking article, or from your own draft, and paste it into the box. Then click Analyze.
3. Look at the "Entities" tab
There are other tabs here that do some cool stuff, but what we care about in this process is "Entities". This tab shows the list of entities detected in the text, along with the importance of the term in the content (with 1 being the most important) and the category of the term (price, number, person, consumer good, etc.)
4. Compare top content to your own
Paste your content into the tool and compare it to a top-performing article. Do you have the same level of detail? Are you covering the same types of concepts?
If important entities are missing or low in salience, it’s a sign your content may be too shallow or off-topic and you may need to rewrite with more depth and relevance.
Wrapping it up
Like I said at the start, this isn’t a call to throw out keyword research. It's still essential to know what people are searching for, how they phrase their questions, and where the opportunity lies.
What this is about is going one layer deeper. Once you know the right keyword to target, writing with entity salience in mind helps ensure your content doesn’t just match the query, but actually answers it with depth, relevance, and credibility.
Both matter. Keywords tell you what to write about. Relevant, salient entities help prove you know what you’re talking about.
Want to learn more about optimizing for LLMs?
If you're working on a brand or site that needs to show up in large language models, I wrote a separate guide on evaluating and optimizing brand visibility in LLMs that incorporates this process.