Speaker Hub uses necessary cookies for secure accounts. With your permission, first party analytics help us improve signup and onboarding. We do not use advertising or cross site tracking cookies. Read the Cookie Policy.
What actually gets you quoted by AI, and what does not · Speaker Hub
What actually gets you quoted by AI, and what does not
Google deleted FAQ rich results in 2026, and it never read llms.txt at all. Here is what survives once you strip out the advice that stopped being true.
•6 min read
Advice about getting cited by AI ages badly, and most of what circulates is a blog post repeating another blog post. Two of the most commonly recommended tactics are worth checking directly, because the primary sources say something different from the advice.
FAQ schema no longer earns you a rich result
For years the standard recommendation was to add FAQPage markup and collect the expanded question and answer boxes in search results.
That is finished. Google restricted FAQ rich results to authoritative government and health sites in August 2023, then removed the feature outright. Google's own documentation change log records a deprecation notice added on 8 May 2026 saying the feature would no longer appear in Search from 7 May 2026, followed by the removal of the FAQ rich result documentation entirely on 15 June 2026. Search Console stopped reporting on it in January 2026.
So if you are adding FAQ markup because you were promised those boxes, the boxes are gone.
Keep marking up questions you genuinely answer, because the markup still labels a question and its answer unambiguously for anything else parsing the page. Just do not expect it to change how you look in Google, and do not let anyone sell you an FAQ schema audit on that basis.
llms.txt is a proposal, not a channel
The pitch is appealing. Publish a plain text file at your domain root summarizing who you are and what you speak on, and AI crawlers will read it directly.
The llms.txt proposal is real and the format is sensible. Adoption is the honest problem. It is a community proposal, not a standard that search or answer engines have agreed to support, and Google has not stated that any of its Search systems read it. Support across AI products is uneven and concentrated mostly in developer tooling.
None of that is an argument against publishing one. It takes ten minutes and it cannot hurt you. It is an argument against treating it as the reason you are or are not getting cited, and against paying anyone to install one for you.
What actually holds up
Strip out the tactics that depend on a feature being switched on somewhere, and what remains is unglamorous.
Say who you are in a format built for saying it.Person markup with your name, a job title, the topics you cover, and sameAs links to your LinkedIn and other profiles. This is the highest value single change, because sameAs is what lets a machine connect the Jane Doe on your site to the Jane Doe on a conference programme and treat them as one person rather than three strangers.
Spell your name the same way everywhere. If you are Jane Doe on your site, Jane A. Doe on LinkedIn, and Janie Doe in a conference bio, you have made that resolution harder for no reason.
Answer the question in the first sentence. Answer engines lift short self contained answers. If a page addresses a question, put the direct answer first and the reasoning after it. Three paragraphs of setup before the answer means the answer never gets quoted.
Publish transcripts under your talk videos. A video is close to opaque. The transcript beneath it is thousands of words of indexable, quotable text on exactly the topics you want to be known for. This is the cheapest large win available to most speakers, because the content already exists.
Get cited somewhere that is not your own website. A podcast appearance with real show notes, a guest post, or a quote in an industry publication carries weight that another page on your own domain does not. Google's guidance on creating helpful content is the closest thing to a primary source on what this rewards.
How to read advice in this area
Search and answer engine advice has an unusually short shelf life, and almost none of it gets retracted when it stops being true. The FAQ schema recommendation is still being given confidently in 2026, more than a year after the feature it promised was switched off.
One habit fixes most of it. Before acting on a tactic, find the primary source: Google's own documentation, the schema.org definition, the spec itself. If the only citation anyone can produce is another blog post, treat it as folklore until proven otherwise.
Not for Google rich results. Google added a deprecation notice on 8 May 2026, stopped showing the FAQ rich result in Search from 7 May 2026, and removed the feature's documentation on 15 June 2026. Marking up genuine questions still labels them clearly for other systems that parse structured data, so it is not harmful, but any advice promising you a rich result from it is out of date.
Should I publish an llms.txt file?
It costs about ten minutes, so there is little reason not to. Set your expectations honestly though. It is a community proposal rather than a standard any search engine has committed to, and Google has not said that any of its Search systems read it. Do it because it is cheap, not because it is guaranteed to work.
What is the single highest value thing to add?
Person markup with sameAs links to your other profiles, on a page that also states in plain prose what you speak about. It is the one change that turns your name from a string of text into an entity a machine can connect across the web.
The bio that impresses a programme chair and the bio a search engine can use are not the same paragraph. You need both, and most speakers write neither.
