What Will ‘SEO for Chatbots’ Look Like?
AI chatbots are slowly becoming a major way people get information. How do you get them to pay attention to you?
How do AI models like ChatGPT decide what information to point to when serving you answers? The answer isn’t well understood, but there is massive interest in changing that. Now that OpenAI has revealed it has more than 200 million active monthly users, it’s clear that getting chatbots to prioritize YOUR information will put it in front of a lot of eyeballs.
So what does “SEO for chatbots” look like? That was one of the big subjects I tackled during a panel discussion on AI and media put on by Hotwire Global on Wednesday night. In front of a packed room in New York’s FiDi neighborhood, I spoke with Hotwire’s Joanne Denyeau and Lauren Macdonald about how newsrooms are adapting to AI as well as journalists’ deep skepticism about AI, in addition to exploring how PR pros can start to think about the new frontier of chat and AI summarization.
More on that shortly, but I’m thrilled to report I’ll be returning to Fox Business tomorrow morning to talk about AI in the real world. Watch for my segment at about 8:45 a.m. Eastern Time to learn about how AI is slowly becoming your news anchor, doctor, and manicurist(!).
I was also recently on the Backstory on Marketing & AI Podcast, where I chatted with host Guy Powell about many of the chief concerns marketers have about AI: privacy, hallucinations, and is the return on investing in AI worth it? It was an expansive discussion, and Guy is a great host — please give it a listen if you have a minute.
And finally, I’m sitting on a plane as I write this, about to touch down in Atlanta for the Online News Association (ONA) conference. If you’re there too, I’d love to meet. Reaching me is as easy as replying to this email.
OK, a quick reminder about our upcoming AI classes, and then let’s talk about SEO for AI.
Learn How to Use AI for Journalism and PR
The fall is a great time to sharpen your AI skills as you prepare for big, year-end projects. The Media Copilot offers virtual classes tailored to media professionals — journalists and PR specialists — who want to stay ahead of the curve.
What you can expect from our classes:
Live, interactive sessions
Complete training on everything from basic prompts to specialized tools, all in a single afternoon
Upcoming classes:
AI Quick Start (Sept. 25)
Duration: 1 hour
Cost: $80 per person
Focus: AI fundamentals and efficient prompting techniques
AI Fundamentals (Oct. 4)
Duration: 3 hours
Cost: $375 with discount code AIFALL (regular price $750)
Focus: Advanced prompting, ethical AI use, and a handpicked toolkit for your daily workflow
Secure your spot, build your skills, and elevate your impact.
Chatbot Engine Optimization: The New SEO Frontier
If the large language models (LLMs) that power AI services are good at anything, it’s summarization. Ask a general-purpose chatbot to tell you the causes of the French Revolution, how to bake the perfect peanut butter brownie, or — if it can browse the web — what the latest polls say about who is likely to win the presidential election in November, and it’ll spit back a paragraph or three that breaks down the most important takeaways.
This fundamentally reduces the friction between you and the information you’re seeking. With traditional search, you’d get a set of links, and it would be up to you to click on them, read the content, and apply your own judgment to surmise what you’re seeing. With AI, I think Google says is best with its AI Overviews: “Let Google do the Googling for you.”
Ultimately, this is why AI and chat will inevitably become a significant part of how we get information. Even if OpenAI hadn’t recently told us that its users doubled to 200 million in the past year, we can plainly see that AI summarization is spreading: several publications, including USA Today and News Corp Australia, are doing it within their platforms, and it’s a main draw of AI-based aggregators like Particle. Gartner has predicted a 25% drop in traditional search traffic by 2026 due to the rise of chatbots and summaries like Google’s AI Overviews.
So chat is here to stay. The inevitable question then becomes: how can I get the chatbot to summarize MY information, and not someone else’s? How can I optimize my content for AI?
The short answer: no one really knows, at least not yet. Why a chatbot chooses one particular set of facts over another to answer a query is the result of a number of factors, and a lot depends on the specific tool your using.
Certainly, the same rules that govern traditional SEO apply to AI. You should prioritize unique facts and data, makes sure your sources are authoritative, and keep the information accurate and up to date. Internal linking and external sites linking to you couldn’t hurt.
Search vs. Chat vs. Answer Engines
But things get fuzzy pretty quickly. AI chatbots often give different answers to the exact same query, so keywords — a staple of SEO — aren’t as important. The influence of something like Google Page Rank is also unclear, especially since some services, such as Perplexity, deliberately seek out a diversity of sources to produce answers, not just the sites that would appear at the top of Google.
I’ll stop to make an important distinction: There’s a difference between an AI chatbot summarizing knowledge in its training data, and summarizing websites that the chatbot browsed on the web. Although the end result — a short summary of the information — is the same, that summary is created in a different way: Rather than reading patterns in its own data, the chatbot is applying its language abilities to data from the outside.
It's a subtle difference to the user, but it’s important, because it speaks to attribution. Put simply, when asked to summarize information in its training data, the AI is taking into account many sources, so it’s difficult for it to give you a singular link for any particular fact. When it does so, it’s offered more as “check this out for further reading,” as opposed to “here’s exactly where I learned this.”
For AI-powered search engines — such as Perplexity, SearchGPT, or Google’s AI Overviews — the AI is browsing the web and summarizing what it found. In this case, there is a clear source for each piece of information, so the link is more meaningful. This is where anyone working to decipher “SEO for AI” should focus, since linking and attribution is the norm. In addition, the licensing deals OpenAI has been making with publishers include metered payouts, measured by how often their content is used to answer queries in SearchGPT.
This all speaks to “why do this at all?” With AI-powered search, any optimization efforts have the potential to pay off, both with referral traffic and revenue.
Step 1: Understanding
So how can you measure the set of queries that users will use to surface your data? It’s unclear exactly what data is being shared with publishers, but presumably they get some guidance. Some third-party services — notably Hotwire’s GAIO — are starting to scratch the surface of the black box, measuring which sources are cited in AI search engines from broad set of queries.
Another way to understand how people are searching in chatbots, particularly from your own audience, is to introduce chat on your own website. Chat is fundamentally different from traditional search: users tend to converse with longer queries and are much more specific that simple one- or two-word topics. It also has the potential to keep users on your site longer.
Moreover, by definition the people on your platform are your most loyal readers, and it stands to reason they’re typing similar queries when in Perplexity or ChatGPT. By optimizing your content strategy around those queries, you can take an important step to making your content “AI-ready.”
The Media Copilot can help with that. Alongside our tech partners, we can introduce a chatbot for your site, trained on your content. It’s not the end of “SEO for AI,” but by introducing this new way to interact with your most loyal readers, you can start down the path to mastering our future AI-mediated ecosystem. Hit me up if you’d like to learn more.
The Chatbox, with quick takes on the most recent headlines in AI, will return next week.