The question every PR professional, comms expert and content marketer is asking right now is: How can we increase the chances of brands appearing in AI LLM chat tools like ChatGPT, Google Gemini, Perplexity, and such?
Two interesting recent studies throw some light on this.
1. What Content Works Well in LLMs? (Kevin Indig)
This study by top growth marketer Kevin Indig suggests that the key to getting mentioned in AI chatbots and answer engines is brand-building and creating content that covers topics in greater depth. It’s based on an analysis of 7,000 citations on ChatGPT, Perplexity and AI Overviews across 1,600 URLs from content-heavy websites.
Since AI Chatbots use RAG (Retrieval Augmented Generation) to balance their answers with results from Google and Bing, one of Kevin’s underlying goals was to find out if classic SEO ranking factors are also influencing AI Chatbot mentions and citations (where citations are mentions that include a link to the brand’s website).
Key findings
And surprisingly, those classic search ranking factors don’t seem to correlate much with a brand’s AI chatbot/LLM visibility. Even backlinks don’t have much impact! But brand search volume does matter, though. The analysis found a clear correlation between how often people search for your brand on Google and other search engines and how likely it is to be mentioned in AI chatbots.
So, building a brand will help ( and obviously, PR and content play an important role in this). According to Kevin, “early data shows that the most visible brands are digital first and invest heavily in their online presence with content, SEO, reviews, social media and digital advertising.”
Also, content that goes deeper (with more words and sentences) and is easier to read is more likely to be cited. That’s because longer, more in-depth content has a higher chance of answering specific questions during AI interactions.
Finally, not all AI Chatbots mention brands with the same frequency, the research found. While Chat GPT has been adopted most widely and sends the most referral traffic, Perplexity mentions more brands on average in its answers (see chart).
2 How to Get Your Brand in ChatGPT’s Training Data (Seer Interactive)
In this blog Seer Interactive digs into its own and other research to explore how marketers can get their brand included in the foundational training data for LLMs – to increase the chance of brand mentions when users ask relevant questions.
While it’s focused on ChatGPT, the analysis assumes similar rules will apply to other LLMs.
Key findings
Getting included in AI training data relies on your brand being mentioned online in what Seer calls ‘priority sources‘ (a range of publications and platforms) next to the words describing your products/ services AND the related topics your target audience will likely be asking AI answer engines about.
So if your company provides a CRM, your brand name needs to be mentioned in these sources whenever CRMs and related topics are discussed. And crucially, consistent wording surrounding your brand across these sources is crucial (for example, if it’s “a leading supplier of an easy-to-use CRM product for financial services companies, ” these words must be consistently found across the sources.
Seer highlights three separate tiers of priority sources:
- Tier 1 (Critical data sources): Wikipedia (with entries supported by reputable news sources), PR mentions in the publishing partners Open AI uses to train its models (listed here and includes The Guardian, WSJ, Business Insider, ITPro and The Verge) and clear, factual well-structured content on the company website (which should be scrapable by LLMs). Also (surprisingly) press release distribution services also fall under this tier!!
- Tier 2 (Important data sources): Editorial coverage in highly cited and engaged with industry-specific publications, popular and relevant Reddit threads and high-quality long-form content on platforms like Substack.
- Tier 3 (Emerging data sources): With LLMs starting to go multi-modal, Seer believes branded video content on the likes of YouTube and, eventually, content on the major podcast platforms will soon impact foundational training data.
So how can PR help?
Both studies suggest that you can use PR and content to help drive mentions in AI/LLM searches. This includes leaning on PR to build your online brand and maintain an active online presence, especially achieving consistent visibility on a list of trusted media sources. When it comes to content, you need to make sure that any material you produce is easy to read and covers topics thoroughly and comprehensively.