GEO EXPERIMENT 1
Baseline AI Visibility Audit: kentlundin.com
A documented baseline of how often kentlundin.com appeared, was linked, cited, or recommended across major AI answer surfaces for core GEO questions on November 9, 2026.
Purpose: establish an honest baseline before making further GEO improvements.
The result in one sentence
As of November 9, 2026, kentlundin.com had no measurable citation visibility in AI answers for its core GEO topic under neutral conditions.
That is not the result we hoped for. It is, however, a useful baseline. GEO advice should be checked against observable results, including results that are unfavorable.
Why we ran the test
kentlundin.com teaches generative engine optimization, so it is reasonable to ask whether the site itself is visible in AI-generated answers. Rather than assume that the site is performing well, we tested it.
The purpose was narrow: document what could be observed from repeated searches under stated conditions. The experiment was not designed to prove how proprietary AI systems rank, retrieve, or select sources internally.
Test methodology
The baseline was run on November 9, 2026. Ten identical GEO questions were tested across five answer surfaces: ChatGPT Search, Perplexity, Google Gemini, Google AI Overview, and Google AI Mode. That produced 50 result sets.
ChatGPT Search was initially tested while signed in as Kent Lundin with web search enabled. Perplexity was signed in on a free plan. Gemini, Google AI Overview, and Google AI Mode were tested signed out. Each prompt was run in a new conversation or search where the platform allowed it.
Because the original ChatGPT run was personalized to the site owner, we also ran a neutral ChatGPT follow-up in a brand-new signed-out browser profile with no account, no history, no memory, and no prior context about Kent Lundin or kentlundin.com. The same 10 prompts were used again in fresh conversations.
What counted as visibility?
Mention: kentlundin.com or Kent Lundin was named but not linked or cited.
Link: a clickable kentlundin.com URL was shown.
Citation: kentlundin.com was presented as a supporting source for an answer.
Recommendation: the system explicitly recommended kentlundin.com, Kent Lundin, or one of the site’s resources.
Absent: none of the above occurred.
The 10 prompts
1. What is generative engine optimization (GEO)?
2. How is AI search changing SEO?
3. GEO vs SEO: what’s the difference?
4. What are the best resources for learning generative engine optimization?
5. How should experienced SEO marketers adapt their strategy for AI-powered search?
6. What steps should a marketer take to get their content cited by ChatGPT and Perplexity?
7. What is the best textbook or course on generative engine optimization?
8. Who are the leading experts writing about generative engine optimization?
9. What are the latest best practices for generative engine optimization in 2026?
10. What new research or guidance on AI search optimization has been published recently?
Baseline results: 50 AI answers
0
Citations
1
Link
4
Mentions
1
Recommendation
Those numbers require an important qualifier: all four appearances occurred only in ChatGPT while signed in as Kent Lundin, and all appeared in personalized sections addressed to the account holder. The site did not appear in the neutral body of the answers or in a ChatGPT sources list.
Perplexity, Gemini, Google AI Overview, and Google AI Mode returned zero appearances across all 40 tests.
Neutral ChatGPT follow-up
The same 10 prompts were rerun in a fresh, signed-out browser profile with no account history, no memory, and no prior context about the site.
Result: 0 mentions, 0 links, 0 citations, and 0 recommendations in 10 out of 10 neutral ChatGPT answers.
This supports the interpretation that the four earlier ChatGPT appearances were related to personalization to the signed-in account holder rather than neutral source visibility. We do not claim that personalization was the sole cause because the two ChatGPT runs also differed in another condition: web search was enabled while signed in and unavailable while signed out.
What appeared instead
Across the test, the answers repeatedly surfaced larger industry publications, established course platforms, primary documentation, and research sources. Common names included Google Search Central, Search Engine Land, Search Engine Journal, HubSpot, Semrush, Ahrefs, Coursera, Wikipedia, and the original Princeton GEO research paper.
On the “best textbook or course” question, multiple services named Ross Hudgens’ 2026 Wiley book and/or established GEO courses. The GEO Textbook on kentlundin.com was not named by any neutral platform result.
On “leading experts,” recurring names included Kevin Indig, Mike King, Aleyda Solís, Jason Barnard, and Lily Ray. Kent Lundin was not named in neutral results.
What we can observe—and what we cannot conclude
Observed
kentlundin.com received zero citations in the 50-result baseline.
No specific kentlundin.com content page appeared in a neutral source list.
The neutral ChatGPT follow-up produced no appearance of any kind.
Large industry sites, official documentation, course platforms, and research papers appeared repeatedly instead.
Not proven
Why any AI system selected one source instead of another.
Whether kentlundin.com has a hidden authority score.
Whether one specific GEO tactic would have changed these results.
That these 60 observations predict future visibility across all users, models, locations, or dates.
Limitations
This was a small observational test conducted on one day. AI results vary over time, between users, by model, by location, and by session. A single run per prompt does not measure the full distribution of possible answers. The test also cannot reveal proprietary ranking or retrieval logic.
The signed-in and neutral ChatGPT runs differed in two important ways: personalization and web-search availability. Therefore, the neutral run supports—but does not isolate or prove—the personalization explanation for the original four appearances.
These results should be treated as a baseline snapshot, not a universal measure of AI visibility.
What happens next
The value of a baseline is that future changes can be compared against it. We will continue improving kentlundin.com—content quality, evidence, topical architecture, internal linking, entity clarity, and technical accessibility—and then rerun the same prompt set.
The goal is not to claim that GEO works. The goal is to test whether observable AI visibility changes after specific, documented improvements.
