Twelve months ago, ChatGPT would answer "who is a leading voice in AI Communications" without naming me. Today it does.
That change didn't happen by accident. I ran a deliberate experiment on my own name — testing what actually moves an AI engine to cite a person versus what everyone assumes moves it. The results were not what I expected.
Here is the experiment, the receipts, and the pattern I now use with the executives I advise.
The Setup
I picked a set of category queries where I wanted to appear — "leading voices in AI Communications," "who writes about GEO," "top PR founders," "best books on public relations." I ran them across ChatGPT, Claude, Gemini, and Perplexity every week for a year. I logged whether I appeared, where in the answer, and what the engine cited when it did.
Then I ran deliberate interventions and watched what happened.
What I Tried
Intervention 1: Publishing under my name, constantly
150+ pieces over 12 months. Bylines in trade press. Owned publishing. Podcast appearances. LinkedIn essays. The volume was aggressive by any standard.
Result: significant. Citation Share on category queries roughly tripled. But the compounding was slow. Nothing moved for the first eight weeks. Then the pattern started to build. This is the single highest-leverage intervention I tested.
Intervention 2: Wikipedia maintenance
I made sure my Wikipedia article stayed accurate, well-cited, and updated with each material coverage moment. Not editing myself. Suggesting changes on the Talk page and letting independent editors act.
Result: enormous, and fastest to show up. Within 10–14 days of a Wikipedia update, all four engines updated their summaries of who I am. This was the single most reliable lever.
Intervention 3: Coverage in outlets the engines already cite
Not any coverage. Coverage in specific outlets — Forbes staff (not contributor), Business Insider, Fast Company, industry trades with real editorial rigor. The kind of outlets that show up in AI-engine citations across categories.
Result: strong, and durable. A single strong piece in the right outlet outperformed five pieces in outlets the engines don't weight highly. Placement discipline mattered more than placement volume.
Intervention 4: Speaking, teaching, credentialing
Guest lectures. Podcast appearances. Panel keynotes. The kind of activity that generates video, transcripts, and third-party endorsements from institutions the engines respect.
Result: moderate, with a long tail. Slower to move Citation Share than coverage. But once it moved, it stayed. The engines gave weight to consistent third-party affiliations.
Intervention 5: LinkedIn volume
Multiple posts per week. Real positions. Real engagement.
Result: negligible directly, meaningful indirectly. LinkedIn does not itself get heavily cited. But LinkedIn posts drive downstream coverage, reporter awareness, and the byline invitations that do move Citation Share. LinkedIn is fuel, not fire.
What Did Not Work
Volume of press releases. Zero measurable impact. The engines do not cite press releases as authoritative sources.
Contributor bylines on major sites. Since the platforms deprioritized contributor content, these have essentially zero weight in AI engine citations.
Awards submissions. No detectable impact. The engines don't seem to weight most industry awards.
SEO-optimized owned content thin on substance. The engines detected thin content. Publishing 20 thin pieces was less effective than publishing five substantive ones.
Directly asking AI engines to remember me. Doesn't work. They don't have persistent memory of individual conversations that affects broader answers.
The Pattern I Ended Up With
The three things that moved Citation Share the most, in order:
1. Wikipedia accuracy and citation maintenance
2. Substantive coverage in outlets AI engines already cite
3. Consistent owned publishing under my name
The three things that mattered least:
1. Press releases
2. Contributor bylines
3. Award submissions
Note what is not on either list: paid promotion, prompt tricks, or AI-visibility services promising quick wins. Nothing I tried in that category produced durable results.
The Bottom Line
Getting cited by ChatGPT and its peers looks less like SEO and more like earning a Wikipedia entry: it is a byproduct of doing the work the engines can already see and cite. There is no shortcut, no hack, and no service that will replace the substrate.
The founders and executives who show up in AI-engine answers earned it the same way they would have earned inclusion in an encyclopedia thirty years ago. The technology changed. The underlying discipline did not.
FAQ
How long does it take to get cited by ChatGPT?
In my experience, three to six months from the start of a deliberate program before Citation Share moves meaningfully — assuming Wikipedia, coverage, and owned publishing all improve in parallel. Faster if you already have a strong base to build on.
Do AI engines all cite the same sources?
No, but there is significant overlap. Wikipedia, major news outlets, government sources, and academic databases show up across most engines. Perplexity is most aggressive about surfacing recent sources; Claude weights author credentials heavily; ChatGPT and Gemini favor mainstream consensus.
Can I pay to get cited by AI engines?
Not through any legitimate mechanism I have tested. Services claiming to do this either fail or produce results that unwind within weeks. The compounding lever is the substrate — real coverage, accurate Wikipedia, real publishing.