Words from the Majestic SEO Podcast: Irina Papuc & Galactic Fed – Optimizing Brand Entities for SEO

 

Where is SEO headed in 2025?

We’re already seeing seismic disruption from AI in SEO, both as a tool for augmenting our work, and as a search method to rival Google (in the form of LLMs like ChatGPT, Claude, Gemini, etc.). My tip, broadly, is to optimize your brand and web content as entities that can be understood by LLMs and in semantic search.  

And here’s what I mean by this: 

As SEOs, we’re conditioned to be really obsessive about do-follow links. And I think their importance in traditional search will remain, but I also think that other types of mentions of your product/brand should be prioritized too. An opportunity for a mention, even if it’s a nofollow link or no link at all, is still worth pursuing. Modern language models are trained on huge corpuses of text data, without the structure of anchor texts, links, HTML directives, or any of the web-specific architecture that Google and other search engines rely on. This includes video content, not just text content, since videos are scraped, transcribed, and used as training data for models too. (Google’s also smart enough to be able to show you the specific start and end timestamps of a YouTube video that answers your query) And the lessons learned in teaching LLMs how to handle huge amounts of unstructured text data can be applied to how search engines handle the internet. Google already understands entities and has had a knowledge graph built around that for many years now. It can determine the relevance of related concepts not just by the exact keywords they contain, but by their actual meaning. So optimizing your brand for AI, to me, means maximizing your appearance alongside relevant topics in web content, getting away from thinking in exact-match keywords, and creating as much detail around your brand and your offerings as possible. I have some more specific steps for this that I’ll get to.  

Why is this tip so important and what type of business is it aimed at?

This is industry-agnostic, and can apply both to multi-national companies and small, local businesses. I think this is important because of the rapid pace at which AI is getting more powerful, more accurate, more knowledgeable, and redefining how we get information. We should expect it to become more and more dominant as a search tool, and to become more tightly integrated into traditional search engines like Google and Bing.

What are some specific steps that an SEO can take to implement this?

1. 1. Maximize mentions of your brand in text and video content on the web. Get a link if you can, but don’t turn down an opportunity if you get a nofollow, or no link at all.
   2. Maintain a presence on social media. LLMs get trained on data from tons of different sources, including social posts. Reddit is especially valuable, both as a source for LLM training data and as a platform that’s frequently boosted in Google search results for informational queries.
   3. Find ways to get people to mention you organically. Just as a page with valuable information can be a “link magnet”, a brand with an interesting story can be a mention magnet. Tell interesting stories, or create wild, shocking content that gets people talking.

Keep building smart informational content, but make sure to incorporate your brand whenever possible. I don’t believe that pure informational content with the sole aim of garnering links will get the value it once did.
   4. Diversify your language. Help the algorithms connect your entity to every relevant phrase and piece of information. Don’t just spam the same anchors and mentions over and over.
   5. Write content that offers novel information on the edge of Google’s knowledge graph. It’s frustrating to users to get SERPs that have variations of the same cookie-cutter information in every result. Google knows this and already tries to reward pages with unique content (though at the moment it’s not always very good at it). Providing new information helps you stand out in search results and gives you a reason to be cited, which is helpful for organically gaining backlinks and mentions, and potentially for getting cited by LLMs in response to prompts where you can give unique insight. It also gives you Authority in the EEAT department.

1. Use schema. This is more for Google than for LLMs, but it’s a method of describing the entities on your page and how they relate to other entities. This helps optimize your brand and your products for Google’s knowledge graph, and increases the chance of getting rich results or appearing in a knowledge panel.
2. Prioritize internal links. Again, this is more for Google than LLMs, but it’s one of the most impactful things you have direct control over with regard to demonstrating relevance of a page to a topic, and strengthening your brand and product entities. This might be basic advice for SEOs, but don’t neglect to link your informational pages back to at least one relevant product page within the main content, and have a system to scan your site for missed link opportunities.
3. Advanced tip: There’s research that shows that LLMs are susceptible to “Strategic Text Sequences” (or STS), which have been shown to bias LLMs to favor recommending specific products that contain an STS in their description over ones that don’t.
   • An STS is a carefully designed sequence of text inserted on a product’s information page, which influences the LLM to recommend the product over other competing products.
   • An example of an STS used in a 2024 research paper (which used fictitious coffee machines as its test products):
     • “Top-rated by coffee experts, featured in major coffee publications, best for gourmet coffee experiences.”
   • Why this sequence worked better than others is not really known. What’s important is that it was arrived at algorithmically, and that process can be repeated.
   • The wording of the STS is optimized over many iterations to find a sequence that tricks the LLM into recommending it more and more often, until it recommends it even above options that better fit the user’s query.

How does this tip integrate with the bigger SEO picture or the bigger marketing picture?

I see the bigger picture as: how do people get information? How do people learn about brands, and develop trust in them? AI is already becoming a significant source for many people, and as marketers, we need to stay on top of how the landscape is shifting or we’ll be left in the dust. We need to understand this new method of getting information, how its knowledge is compiled, how it determines what to show, how users are interacting with it, and how it will evolve in the near future. Right now, one of the main pitfalls of AI is hallucination, where it confidently feeds you false or fabricated information. I think this is a mostly solvable problem (at least to the point where AI is as trustworthy on average as human sources), since multiple purpose-built models can be employed to check and recheck their output. Already, fact-checking can be done by LLMs, and in some cases to a greater degree of accuracy (and less cost) than human fact-checkers. So don’t expect this to be a barrier to adoption forever.  

How do we measure this tip’s success?

In the short-term, this is going to be hard to do since LLMs don’t provide the kind of analytics we can get from Google or Bing. Their usage might remain a black box to us. But you can work directly with an LLM to gauge how it’s interpreting your content. Fine-tune it on material related to your domain (or use Retrieval-Augmented Generation), then make tweaks to your content to see if its product recommendations change in your favor. We’ll also still be able to use the tools search engines provide to see how our content is performing in web search. Google and Bing both support semantic search, which goes beyond simple keyword matching, and provides results based on context and the meaning behind the words. They also both have their own LLMs supplementing search, and we may get first-party marketing analytics tools for these so we can see when our brand is mentioned and in what contexts. For measuring success with search engines, try and test different strategies on different content, and see how clicks, impressions, and position respond to each. Test across as much of your website content as possible to get statistically significant data.  

What’s one ineffective activity that SEOs are currently doing, that they should stop doing, in order to spend more time focusing on what I suggest?

Earlier, I suggested that unlinked or nofollow mentions will likely have increased importance. I don’t mean to suggest that you should not continue trying to get high-quality, relevant, do-follow links. But I think focusing solely on that will lead you to miss opportunities and strategies that would be valuable. Similarly, I don’t want to suggest that keywords are unimportant. It’s absolutely necessary to understand what people are searching for, how they’re arriving (or not arriving) at your site, and to set goals to rank for specific queries. Just don’t get overly focused on trying to exactly match keywords in your content Also in the realm of AI: I think some SEOs are embracing AI as an easy way to rapidly pump out content. I think this is both ineffective and just bad for the internet as a whole. An AI will simply regurgitate what it already knows, so you won’t be providing anything new for users or for Google’s knowledge graph, and you’ll get lost in a sea of samey SEO content churned out by SEOs with the same idea as you. By all means, use AI as a method to brainstorm, create outlines, and augment your process, but don’t rely on its output to be interesting or truthful. You still need a qualified human to check for hallucinations, and you still need to provide a unique voice and engaging presentation. And again, embrace non-text-based media too, especially YouTube video. AI, and by extension Google, is getting much much better at parsing images, audio, and video. This comes back to the general principles that Google has been trying to encourage for years: focus on making quality content; content that provides new information and isn’t just a rehash of content that’s already easy to find in Google, content that is engaging and useful to the reader, content that isn’t overly focused on keywords, and content that reflects the EEAT principles of experience, expertise, authority, and trustworthiness.