Written by Jonathan Dunnett
5 minute read
In brief: AI tools like ChatGPT and Perplexity build speaker recommendations from different training data and crawl indexes, which means the same booking brief can return completely different results depending on which tool your buyer uses. Visibility requires clarity, consistency, and presence across all the sources these systems pull from.
Many speakers ask: why doesn’t AI recommend me as a speaker? The answer isn’t about how good you are on stage. It’s about something more specific and more fixable.
This post is the third in a series on AI visibility for keynote speakers. Issue one covered website structure. Issue two covered identity consistency across your online properties. This one is about what actually happens when the system runs the search.
What AI Speaker Recommendations Actually Look Like
I’ve been running an experiment I call Speaker Roulette. The premise: generate a realistic keynote booking scenario — real city, real topic, real budget — and drop it into ChatGPT and Perplexity to see who comes back. I “create” the brief, just like an event planner might actually type it.
A recent session used this scenario: a mid-size conference with a $28,000 speaker budget, focused on resilience and change management.
ChatGPT returned six names, and Perplexity returned three: nine speakers across two tools, with zero overlap.
Not one speaker appeared on both lists with the same brief. Two of the most widely used AI tools in the industry right now are showing entirely different outputs: and that’s to be expected.
Why the Two Lists Were Completely Different
ChatGPT and Perplexity are drawing from different training data, different crawl indexes, and different signals about who is credible on a given topic.
Each tool is building its own version of “who speaks on this” based on what it has access to, and recommending from that version. The same search, run on a different platform, returns a completely different set of names.
For speakers, this has a direct implication: your visibility isn’t just about being good. It’s about being understandable and recommendable to the specific system that happens to be running the search on the day your potential client goes looking.
Which raises a question worth sitting with: which AI tools are your buyers actually using?
The Pattern Nobody Is Talking About Openly
When I looked at the two lists from that session side by side, the zero overlap wasn’t the only thing that stood out.
The composition of the lists was different in ways that go beyond topic expertise. ChatGPT returned six names: all men. Perplexity returned three names, two of them women.
Same brief. Different training data. Different results.
Jennifer Moss, a speaker who showed up in an earlier Speaker Roulette session, commented on the video: “It was interesting that ChatGPT chose all men and Perplexity chose two women and then generalized… I need to look better at how each one searches now.”
Tiana Sanchez, another speaker who watched the results, noted: “I’m disheartened that no women were found in the first search of five speakers.”
We know algorithmic bias exists. We’ve seen it in hiring tools, in image generation, in search results. There’s no reason to assume AI speaker recommendations are immune to it. The training data reflects what was written about, cited, and amplified historically — and historically, that hasn’t always been equitable.
I’m not drawing a definitive conclusion from one session. But it’s worth watching, worth running your own tests, and worth understanding that the system surfacing you — or not surfacing you — isn’t neutral.
Why AI Doesn’t Recommend Me as a Speaker — Even Well-Known Ones
This is the question I hear most often, and the answer surprises people. The speakers who don’t show up in these sessions aren’t all unknown. Some have sold hundreds of thousands of books. Some have very healthy businesses built entirely on referrals and bureau relationships.
They’re invisible to AI not because they’re not good enough, but because the AI can’t resolve who they are with enough confidence to stake a recommendation on them.
A confused AI doesn’t hedge. It just moves on to someone it is sure about.
What creates that confusion? Usually one or more of these:
An online presence that describes too many things. Five different topics, five different audiences, five different versions of the same person. The AI tries to build a picture and can’t.
A bureau listing that’s out of date. Written three years ago, before you pivoted your focus. It describes a version of you that no longer matches your website — and the AI sees the contradiction.
A website that doesn’t clearly signal that speaking is primary. “Speaker” buried in a list of services. No dedicated speaking page. No clear repetition of the topic you’re known for.
Third-party content that doesn’t reinforce your lane. Podcast appearances, articles, interviews — they exist, but they cover five different topics instead of building a consistent signal around the one thing you want to be recommended for.
Each one of those is a signal. Or a contradiction. The AI is reading all of them, across all of your properties, and building a picture. Whether that picture generates a confident recommendation depends entirely on what you’ve given it to work with.
How AI Decides Which Speakers to Recommend
Understanding how these systems actually work helps explain what you can do about it.
Large language models like ChatGPT and Perplexity don’t search the web in real time the way Google does (though some have web access). They’re drawing on training data, crawled content, and signals about authority and credibility that have been built up over time.
When someone asks for a speaker recommendation, the model is essentially asking itself: who do I know that fits this brief, and am I confident enough in that answer to recommend them?
Confidence comes from:
- Clarity — is this person’s topic and audience unambiguous across multiple sources?
- Consistency — does their website, LinkedIn, bureau listing, and third-party coverage all tell the same story?
- Presence — are there enough sources referencing this person in this context for the model to feel certain? Often, “bigger and more credible” places matter. Meaning, if you’re cited by CBS news, that means more than Joe Blogger.
Missing any one of these creates doubt, and in some cases, the model moves on to someone else it can recommend confidently.
What This Means for Your AI Visibility
Understanding why AI doesn’t recommend me as a speaker — or you, for that matter — starts with how these systems build confidence in a recommendation. The speakers who are showing up in these sessions aren’t necessarily the biggest names, but they tend to have one thing in common: there’s no ambiguity about what they do and who they do it for.
It’s genuinely hard to be a “wellness and resiliency speaker” in 2026 and expect AI to know what to do with you. The more specific your lane, the more consistently you signal it across every property, the more understandable you become to the systems your buyers are starting to use.
Clarity is a competitive advantage right now: because most speakers aren’t thinking about this yet.
Here’s where to start:
Audit your bureau listing. When was it last updated? Does it reflect your current positioning, your current topics, your current audiences? If not, it’s actively working against you.
Check your LinkedIn profile. Is “speaker” the first thing someone reads, or is it buried under “author, coach, consultant, and facilitator”? The model is reading your headline.
Look at your third-party footprint. The podcast appearances, the articles, the interviews — do they reinforce a consistent topic, or do they scatter your signal across ten different subjects?
Own a lane clearly. Not “leadership and culture and change and resilience.” One owned topic, signaled consistently, across every property you control.
Watch Speaker Roulette
The full session — the city, the scenario, the complete lists from both tools — is in the video below. Seeing the zero overlap play out in real time makes the point in a way that a list of best practices can’t.
Want to Know Where You Stand?
If you want to know what these tools actually return when someone searches for a speaker like you (and what to do about it), you can book a strategy call here. I’d love to take a look.
This is exactly the work I do with speakers in the Does AI Recommend You? workshop — a structured process for auditing and improving your AI visibility, in a small group setting with speakers who are serious about building a more predictable pipeline.
This is part three of a series on AI visibility for keynote speakers. Part one: Before AI can recommend you, it has to understand you
Jonathan Dunnett helps keynote speakers get booked on purpose. He is a Council of Competitive Intelligence Fellow and brings systematic, intelligence-driven frameworks to speaker business development. Serendipity isn’t a strategy.
Frequently Asked Questions
Why doesn’t AI recommend me as a speaker?
AI tools build speaker recommendations from training data, crawled content, and authority signals. If your online presence describes too many topics, contains contradictions between properties, or doesn’t clearly and repeatedly signal that speaking is a primary activity, the AI can’t build a confident picture of who you are — and a confused AI moves on to someone it is sure about.
How does AI decide which speakers to recommend?
AI tools look for clarity, consistency, and presence. Clarity means your topic and audience are unambiguous. Consistency means your website, LinkedIn, bureau listing, and third-party coverage all tell the same story. Presence means there are enough sources referencing you in the right context for the model to recommend you with confidence.
Do ChatGPT and Perplexity recommend the same speakers?
Not necessarily. In a live test using the same booking brief, ChatGPT returned six speakers and Perplexity returned three — with zero overlap. Each tool draws from different training data and crawl indexes, which means your visibility on one platform doesn’t guarantee visibility on another.
How do event planners use AI to find keynote speakers?
Increasingly, event planners are using AI tools like ChatGPT and Perplexity to generate speaker recommendations directly from a brief — specifying topic, audience, budget, and location. The tools return a shortlist of names, which the planner then researches further. If you’re not in that initial list, the conversation starts without you.
How do I improve my AI visibility as a keynote speaker?
Start with your website — make sure it clearly and repeatedly signals that you are a keynote speaker on a specific topic. Then audit your bureau listing for outdated positioning, check your LinkedIn headline, and review your third-party content to ensure it reinforces your lane rather than scattering your signal.
Ready to Work on Your Speaker Positioning Strategy?
I’m running a small group training workshop for keynote speakers who want to build a more systematic, predictable approach to getting booked. We’ll cover AI visibility, identity consistency, website fundamentals, and outreach strategy, with a small enough group that there’s room to work through your specific situation.