GEO for Speakers:
The Shortlist Is Now a Prompt

Quick Answer:

GEO for speakers is the practice of structuring a keynote speaker's public track record — bios, talk pages, event coverage, third-party mentions — so AI assistants can extract, corroborate, and cite it when an organizer asks for recommendations. The working method: run the Shortlist Test (8–10 organizer-style prompts across ChatGPT, Gemini, and Google AI Mode, scoring mentions and citations separately), then close the gaps layer by layer: Discoverability, Clarity, Authority, Trust. The lever most speakers miss is corroboration — AI engines trust what independent sites say about you more than what your own site says.

The shortlist moved inside the model

The volume shift is not subtle. AI search platforms received 27.4 billion visits in Q1 2026, up 42.8% year over year, with ChatGPT alone accounting for over 16.8 billion, according to Wix Studio's AI Search Lab. The event industry is inside that curve, not watching it: the Amex GBT Global Meetings & Events Forecast reports half of industry professionals now integrate AI throughout event planning and execution, and Bizzabo's 2026 research found 95% of event professionals expect AI use in events to increase.

Meanwhile the old discovery surface is quietly shrinking. When a Google search triggers an AI Overview, Pew Research Center found users click a traditional result in only 8% of cases versus 15% without one — and click a source inside the AI summary just 1% of the time. Even in classic search, the demand was never where speakers think: in Semrush's US database as of July 2026, “what is a keynote speaker” draws 1,000 searches a month, while “how to find a keynote speaker” draws 110 and “how to choose a keynote speaker” just 70. Discovery queries were always low-volume, high-stakes — exactly the query type that moves fastest into a private AI chat where no one can see it happen.

For a speaker this changes the failure mode. Under SEO, being invisible meant ranking on page two — recoverable with effort. Under AI-assisted shortlisting, being invisible means the organizer's first answer names six competitors and the conversation proceeds without you. There is no page two in a chat answer.

Run the Shortlist Test

The Shortlist Test is a monthly audit: 8–10 organizer-style prompts, run on three AI platforms, scoring mentions and citations separately. It takes under an hour and replaces guesswork with a baseline. The prompts must be the ones a real organizer would type, not vanity queries:

• “Best keynote speakers on [your topic] for 2026”
• “[Topic] speaker for a CMO summit in [your region]”
• “Who speaks about [your topic] in [language]?”
• “Alternatives to [the famous name in your category] for a corporate event”
• “Keynote speaker who covers [topic A] and [topic B] with original research”

Score each answer twice, because being mentioned and being cited are different mechanisms. Semrush's AI Visibility Index 2026 — a study of 126 million US AI-search prompts across 22 verticals, January to April 2026 — found the overlap between the brands a platform mentions and the sources it cites ranges from 64% on Google AI Overviews down to 30% on Gemini. You can be named without any of your pages being read, and quoted without being recommended. Track both numbers; they respond to different work. I unpacked the distinction in Why Citation Authority Matters More Than Rankings in AI Search.

Speaking from my own audit: I run a version of this test on my name and category every month. In June 2026 I appeared in 18 of 30 category prompts — strong on branded and Spanish-language queries, absent from most “best GEO experts” listicle-style answers. That gap is not an insult; it is a work order. It tells me exactly which prompt classes need corroboration I don't yet have.

One platform is not a strategy

The instinct after a bad Shortlist Test is to “optimize for ChatGPT.” The data says that covers less of the map than it appears to. Profound's analysis of 680 million citations (August 2024–June 2025) found only 11% of domains cited by ChatGPT are also cited by Perplexity — and a separate 2026 analysis of 118,000 AI answers (Qwairy) found only 11% of cited domains appear on more than one platform. The engines are reading almost disjoint webs.

Platform behavior diverges on volume, too. Per the AI Visibility Index, ChatGPT averages 15.4 citations per answer; Gemini averages 3.3. And each engine keeps its own source diet: ChatGPT over-indexes on Reddit, Gemini on Wikipedia, and Google's AI Mode and AI Overviews on YouTube. Across platforms, AI answers cite Wikipedia 4.3 times for every brand mention. For a speaker, the translation is concrete: your YouTube keynote recordings work on Google's surfaces, community threads discussing your talk work on ChatGPT, and a well-sourced Wikipedia or Wikidata presence works on Gemini. Different rooms, different doors.

The four layers, applied to a speaker

The AI Visibility Index frames brand visibility as four stacked layers — Discoverability, Clarity, Authority, Trust. They map onto a speaker's presence cleanly:

• Discoverability — can the machine find a consistent you? One canonical name spelling, one job title, one affiliation string everywhere: site, LinkedIn, event pages, podcast show notes. Person schema with sameAs links tying your profiles into one entity. An llms.txt file stating who you are in plain text. Inconsistent bios don't read as variety to a model; they read as different people.
• Clarity — can the machine quote you cleanly? Talk titles that name their claim (“Mentions vs Citations: The Two Metrics Every CMO Conflates” extracts; “Winning Tomorrow” doesn't). A speaker page that answers, in extractable sentences: topics, formats, languages, regions, proof. FAQ blocks phrased the way organizers ask — the same question-first structure I described in What Is Generative Engine Optimization?
• Authority — does anyone else say it? This is the layer speakers skip, and the one the engines weigh most. Event recap posts naming you, conference speaker pages that stay live, podcast appearances, industry-press quotes — every one is a corroborating domain. The Princeton/IIT GEO study (SIGKDD 2024) measured a 22–41% visibility improvement in generative answers from optimization techniques — and citation-adding methods beat formatting tricks. Ask every organizer for two artifacts: a live session page and a recap mention. Ten events a year compounds into a corroboration graph no competitor copies quickly.
• Trust — is it fresh and verifiable? A ConvertMate analysis of 10,000+ domains found 76.4% of ChatGPT citations come from content updated within the previous 30 days. A speaker page listing 2024 events as “upcoming” is an anti-signal. Dated testimonials with full names and roles, a maintained past-events list, and stats with sources on every claim close the loop.

The stage-to-citation loop: what I do after every talk

Stage time is the raw material AI never sees unless you convert it. My loop, after 50+ events a year: within two weeks of a talk, the core argument becomes a blog post carrying the slide data — sourced, dated, question-structured. The organizer gets a quotable summary for their recap page, which links back. The Q&A questions feed the FAQ blocks, phrased exactly as real buyers asked them — the method from The Q&A Is the Research. The recording goes to YouTube with a claim-naming title, because that is the source diet of Google's AI surfaces.

Each stage appearance thus produces four citable artifacts on three source diets — my domain, the organizer's domain, YouTube — every month. That cadence is also what the freshness data demands: under the 30-day citation window ConvertMate measured, a speaker site updated quarterly is a speaker site that is structurally stale for two of every three months.

What GEO won't do for a speaker

Honesty section. GEO does not fix a weak talk — it amplifies whatever record exists, and if the record is thin, amplification produces nothing. It does not replace bureaus and referrals: high-fee bookings still close on human trust, and a bureau profile is itself a corroborating domain, not a competitor to this work. And it is not a traffic play — with discovery queries in the dozens per month, nobody should expect analytics dashboards to light up. The metric that moves is quieter and worth more: the share of organizer-style prompts in which the model knows you exist. Speaker booking was always a zero-click market — decisions made on referrals and reputation, invisible to analytics. AI didn't create that dynamic. It just moved the referral into the prompt window — and made it auditable for the first time.

If you book speakers rather than compete as one, the same mechanics run in reverse — I wrote the organizer's side in How to Pick an AI Keynote Speaker: treat the model's shortlist as a starting pool with a recency bias, not a verdict.

Frequently Asked Questions

What is GEO for speakers?

GEO (Generative Engine Optimization) for speakers is the practice of structuring a keynote speaker's public track record — bios, talk pages, event coverage, and third-party mentions — so AI assistants like ChatGPT, Gemini, and Google AI Overviews can extract, corroborate, and cite it when an organizer asks for speaker recommendations. Where speaker SEO optimizes one website to rank, GEO optimizes an entire corroboration footprint to be quoted.

Do event organizers actually use AI to find speakers?

Adoption data says yes, and rising. The Amex GBT Global Meetings & Events Forecast reports half of industry professionals now integrate AI throughout event planning and execution, and Bizzabo's 2026 research found 95% of event professionals expect AI use in events to increase. AI search overall reached 27.4 billion visits in Q1 2026, up 42.8% year over year — speaker research is part of the shift.

How do I check whether ChatGPT recommends me as a speaker?

Run the Shortlist Test: write 8–10 prompts a real organizer would use (“best keynote speakers on [topic] 2026”, “[topic] speaker for a CMO summit in [region]”), run each on ChatGPT, Gemini, and Google AI Mode, and score mentions and citations separately. Repeat monthly — platform answers shift as their sources refresh, so the trend matters more than any single run.

What is the difference between speaker SEO and speaker GEO?

Speaker SEO aims to make your own website rank for queries like “keynote speaker + topic”. Speaker GEO aims to make AI assistants name you and cite pages about you — and those pages are mostly not yours. With only 11% of ChatGPT's cited domains also cited by Perplexity (Profound, 680M citations), GEO means building corroboration across many independent sites rather than polishing one.

Why doesn't my speaker website alone make me visible to AI?

Because AI assistants weigh independent corroboration over self-description. Semrush's AI Visibility Index 2026 shows each platform runs a different source diet — ChatGPT leans on Reddit, Gemini on Wikipedia, Google's AI surfaces on YouTube — and AI answers cite Wikipedia 4.3 times per brand mention. Your site states a claim once, on a domain you control; the engines look for the same claim on domains you don't.

How long until GEO changes show up in AI answers?

Freshness cycles are short — 76.4% of ChatGPT citations come from content updated within the previous 30 days (ConvertMate, 10,000+ domains) — but structural work like schema, consistent bios, and new third-party coverage typically needs one to three months of re-crawling before answers shift. Re-run the Shortlist Test monthly and treat the trend as the metric.