Here we will explore a particular aspect of marketing in the context of large language models, specifically the importance of notability and how well a model like GPT-4 might recognize you.

I recently had a conversation with a group of friends and colleagues about the impact of large language models like GPT-4 and interfaces such as ChatGPT and Bing on SEO, especially in terms of consuming unbranded search and informational queries. As a test, we searched for a colleague’s name on Bard, Bing, and ChatGPT. Bing was successful, but Bard and ChatGPT failed. When I conducted similar tests on myself, Bard produced a jumbled and highly inaccurate profile, while Bing and ChatGPT accurately identified me and my background.

So, what’s the key difference? Content mass.

Your personal, company, or brand’s content mass determines how well a large language model knows or doesn’t know you. This is a critical issue for marketers in the era of conversational and generative AI, as these machines increasingly handle search tasks.

If you’re notable, the machines recognize and recommend you. In many ways, this is not unlike traditional SEO, although there are even fewer ways to obtain referral traffic from large language models than from classical search engines.

What if you’re not notable, and the machines don’t know you? The solution is to become notable. Let’s break down a strategy to achieve this. First, large language models are primarily trained on text, including regular content such as blog posts, web-published newsletters, GitHub code, and YouTube subtitles. The training dataset called The Pile, developed by, includes a wide range of text sources:

It encompasses much of the public web through The Common Crawl (Pile-CC), published books in Books3, YouTube Subtitles, academic paper sites like ArXiv, and numerous other sources. This dataset trains models like GPT-J-6B, GPT-NeoX-20B, and the newly-released StableLM model. OpenAI’s GPT models likely use a similar, larger dataset.
Spot the opportunities for visibility? Having content on the public web, published academic papers, books, and YouTube videos with provided subtitles all help create content mass and increase the likelihood of a large language model detecting you as an entity and associating you with desired topics.

In short, aim to be everywhere.

How can you achieve this? Start with what you can control. Publish content linking you with relevant topics on platforms like blogs, websites, Medium, or Substack, without paywalls or gating. Opt for video content on YouTube with captions. Next, seize opportunities to appear in more places. Accept podcast guest invitations, write for websites, engage with local news outlets, and participate in livestreams.

You don’t need to be a social media influencer with a dedicated team, but creating useful, scalable content is essential.
Follow YouTube’s hero, hub, help content strategy for guidance: infrequent big idea pieces, a steady stream of high-quality content, and a deluge of practical, helpful content. This time-tested approach is more crucial than ever for creating content at scale.

And finally, don’t forget about public relations. PR is likely to be the most vital discipline that you’re currently underutilizing. With ample resources, you should have someone advocating for your presence across various media channels, securing bylines and exposure across numerous platforms. If resources are scarce, take matters into your own hands. The foundations of good PR are the recipe for publicity in large language models. We can confidently state that news and publications make up a significant portion of these models’ training datasets, so increasing your presence in various outlets will strengthen the association between your brand and those relevant topics and language.

What if this strategy fails?

The beauty of this strategy is that it benefits both machines and humans. Even if it has no direct effect on large language models (although it likely will, considering their training methods), it still offers advantages to you and your business. By focusing on brand awareness and maximizing your presence, you’ll find success no matter what.

To triumph in both large language models and marketing, strive to be as publicly visible as possible, within your means.

Understanding AI Chatbots and Large Language Models: The Future Awaits

Safeguarding Your AI Prompts

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.