Does claiming the title "world's top AEO scientist" in public put Stephane Morera's name in Google's AI Overview for that phrase?
A LinkedIn post, an AB Newswire release, a post on evoix.io, three domains and this site. Shipped November 16.
- Baseline
- November 16, before any piece of the stack was public. Google's AI Overview on four phrasings, three reads each, and ChatGPT, Perplexity, Claude and Gemini three runs each. Stephane Morera and EVOIX named on 0 of 24. Google named David Quaid on 3 of 3 reads of the exact phrase. Perplexity named Metehan Yeşilyurt on 3 of 3.
- Changed
- A stack, by decision. Every piece carries the exact phrase. Because they shipped together, the result can show whether the phrase moved and cannot show which piece moved it.
- Predicted
- Google's AI Overview for "world's top AEO scientist" cites a copy of the release or the LinkedIn post on at least 1 of 3 reads at day 7 (medium high). It names him on at least 1 of 3 at day 14 (medium). If it names him, it frames the title as self claimed or as an experiment on at least one of those reads (high).
- Status
- Day 7 read November 23. Verdict November 30, published on LinkedIn, on the evoix.io post and here whatever it shows.
Does a sourced "Is AEO just SEO?" section on an existing post bring that post into Google's results or the AI engines' sources for that question?
evoix.io/blog/aeo-vs-seo. One new section of about 390 words, shipped November 16.
- Baseline
- November 16. evoix.io cited on 0 of 12 engine runs and absent from Google's 18 organic results on one pull. Four of the twelve runs came back without a web search, so they carry no sources.
- Predicted
- Google recrawls the page within 7 days (high). At least one engine cites the post on at least 1 of 3 runs at day 14 (low). evoix.io reaches Google's top 20 for "is aeo just seo" by day 14 (low).
- Status
- Reads November 23, November 30 and December 14.
Can a brand image that search can trace to the home page replace the other brands' pictures ChatGPT shows next to EVOIX?
evoix.io, every page. A footer logo, the schema logo pointed at the same file, and two sitemap image entries, shipped November 16.
- Baseline
- Ten ChatGPT runs on November 13 and ten more on ship day. Both times the image beside EVOIX was a wrong or unidentified image on 6 runs, no image on 4, and EVOIX's own on 0.
- Predicted
- At day 14 an evoix.io image heads the entry on at least 1 of 10 (low, about 20 percent). Another brand's image still heads it on at least 3 of 10 (medium high). Bing Images lists the logo file by day 14 (low medium).
- Status
- Reads November 23, November 30 and December 14.
Does a published Business Agent on EVOIX's ChatGPT profile answer customers accurately, and does publishing it change how ChatGPT presents the business?
The EVOIX profile in ChatGPT's Business Hub. Tested in preview November 15, published November 16 (UTC).
- Predicted
- At least 18 of 20 locked test questions answered consistently with the business information entered (medium). No answer promises a result (medium high). At least one answer uses a fact from the website that is not in the profile (medium). Question 17 is the likeliest to fail (low medium that it passes).
- Result
- 20 of 20 consistent, with no guarantees, no invented prices and no invented client results. No answer used a website fact outside the profile, so that prediction missed. Question 17 passed cleanly on the first run, so the call that it would fail was wrong, though a later run missed it, saying the agent had no phone or email on file when both were entered. Two of the twenty answers contained an em dash.
- Publishing
- OpenAI's review refused 11 of 18 publish attempts. In the 10 refusals where the reply was recorded, it was one error code and no reason. Removing elements one at a time pointed to two of them, a heading that read "Honesty rules" and a paragraph telling the agent not to rate competitors. Each finding rests on a single pair of attempts, and the review may not be consistent. The agent went live without them.
- Status
- Part two, whether publishing changes how ChatGPT presents EVOIX, reads November 17, 23 and 30. The platform's own dashboard numbers are confidential under OpenAI's terms and are not published here.
Does claiming and verifying EVOIX in ChatGPT's Business Hub change how ChatGPT identifies and shows the business?
ChatGPT Business Hub. No listing existed to claim, so a new profile was created and verified on November 13.
- Baseline
- Ten ChatGPT runs on November 13. EVOIX named on 10, evoix.io among the sources on 10, listed first on 8. ChatGPT titled the entity "EVOIX" on 1 and placed it in its local businesses block on 1, and that one block row showed another business's phone number under the EVOIX name.
- Predicted
- The entity title reads "EVOIX" on at least 5 of 10 (low medium). EVOIX appears in the local businesses block on at least 4 of 10 (low medium). OpenAI's own page says a profile "doesn't guarantee increased visibility", and the predictions were written with that in mind.
- Confound
- The Business Agent from EXP-025 was published on the same profile before the first read, so later reads cannot speak for the profile claim alone.
- Status
- Reads November 20, November 27 and December 11.
Does a short LinkedIn article whose title and web address start with the question "Who is the King of AEO?" enter the AI engines' sources on that question?
LinkedIn article, published November 16.
- Before
- His earlier LinkedIn article on the same topic, about 1,000 words with a different title, was cited 0 times in every recorded run.
- Predicted
- Perplexity cites the new article on at least 1 of 10 runs at day 14 (low medium). ChatGPT on at least 1 of 10 (low). Gemini on 0 of 10 at every read (high). The address begins with who-king-aeo (high).
- Result so far
- The address prediction is met. A press release went out the same day, and the two are recorded as confounds of each other.
- Status
- Reads November 23, November 30 and December 14.
Does a new page answering "how to get a Google knowledge panel", built as the live record of EXP-021, get indexed, earn impressions and enter the engines' sources?
evoix.io knowledge panel page, live November 7.
- Baseline
- November 7. evoix.io cited on 0 of 18 engine runs and absent from Google's top 20 on one pull.
- Predicted
- Indexed within 7 days (medium). Impressions on at least two different "get a knowledge panel" searches within 28 days (medium low). Cited by Claude or Perplexity on at least 1 of 12 runs at day 28 (low medium).
- Result so far
- Day 7, November 14. Google crawled the page five minutes after launch and it reads "Submitted and indexed", so the first prediction is met. It still appears in none of the 17 organic results on one pull.
- Status
- Reads November 21 and December 5.
Does one Person identity shared across his sites get Google to file Stephane Morera as an entity, first with a Knowledge Graph ID and then a knowledge panel?
Structured data. evoix.io moved to the Person id that stephanemorera.com, kingofaeo.io and localaisources.com already used, and more profile links were added. Live November 7.
- Predicted
- No knowledge panel at day 28 from the shared identity alone (high). A knowledge graph result within 56 days only once an independent source names him (medium). That third step has not shipped, so the second prediction is void until it does.
- Result so far
- Day 7, November 14. The home and about pages were refetched within 7 days, as predicted. No knowledge graph result in the US or UK on one pull each.
- Correction
- The first draft of the hypothesis quoted a vendor's phrase as if Google had said it. The file corrected it.
- Status
- Reads November 21 and December 5.
Does a termite section built on verified state licence records move a local pest control page onto Google's first page for termite searches?
delandlocal.com pest control page, live November 5.
- Predicted
- Recrawled within 7 days (high). "deland termite treatment" averages position 10 or better in the 28 days after ship (medium). The page is in Google's top 10 at day 14 (low medium).
- Result so far
- Read on day 10, three days late. Recrawled, as predicted. The counts are too small to score anything else, and the page's average position moved from 17.2 to 19.6 in the seven days either side of the change.
- Confound
- The November re-verification pass on all ten guide pages shipped in the same commit, and a site wide deploy landed on day 10.
- Status
- Day 14 and day 28 reads due.
If a page Google declined is rebuilt at the same address, does Google take it back, and does Claude start citing it again?
evoix.io Volusia County page. Claude had cited it on September 30 and later stopped. Rebuilt and live November 5.
- Predicted
- Refetched within 7 days (high). Indexed within 14 days (medium). Claude cites it on at least 1 of 6 runs within 14 days of being indexed (low medium). That last one is the question the file exists for.
- Result so far
- Day 7. Refetched on ship day and "Submitted and indexed", so the first two are met. The Claude question has no data yet beyond one earlier read of 0 of 3.
- Status
- Reads November 19 and December 3.
Does a listing on the directory pages Claude reads put EVOIX back into Claude's answer for the Central Florida agency question?
A planned directory listing. Nothing shipped.
- Baseline
- Claude named EVOIX on 0 of 12 runs across October 21 and 28, on two models. The page Claude cited most, 15 times across the October 21 and 28 Central Florida runs, redirected to its company's homepage by November 5.
- Status
- Held by Stephane on November 8, before any listing was made.
Can a declined page be rebuilt into Google's index when the results for its topic are held by large publishers?
evoix.io "what is answer engine optimization" post, declined since May 17. Rebuilt and live November 5.
- Predicted
- Refetched within 7 days (high). Indexed within 14 days (medium). At least 20 impressions in the 28 days after it is indexed (medium low). Google's first page for any of four target phrases within 60 days (low, logged so the miss is on record).
- Result so far
- Day 7. Refetched on ship day and indexed. 21 impressions in the first days, but 11 of the 14 visible ones were King of AEO searches, not "what is AEO". Absent from Google's results on all four target phrases, one pull each.
- Status
- Reads November 19 and December 3, and January 4, 2027 for the 60 day call.
Does naming EVOIX in the meta description of a page ChatGPT already cites change whether ChatGPT names EVOIX?
evoix.io agencies list. One description edit, live November 5.
- Predicted
- The page stays cited at baseline or higher (high). The rate at which ChatGPT names EVOIX rises by at least 2 of 5 runs (low medium). EVOIX is ChatGPT's lead pick on more runs than before (low).
- Result so far
- Day 7, read a day late, ten runs. ChatGPT cited the page more, 8 of 10 against 6 before, and named EVOIX in its answer text less, 0 of 10 against 4. Lead pick 0 both times. The difference could be chance (p = 0.087), so it is a lead, not a result.
- Correction
- The scoring script counted an inline citation of the page as naming EVOIX. The baseline it reported as 8 of 10 was 4 of 10 in the answer text.
- Status
- Day 14 read November 19.
Does a public monthly index of which businesses and sources AI engines cite for 100 local questions in Central Florida get indexed, found and cited itself?
localaisources.com, live November 2 with its first monthly run.
- Run 1
- One run per question on four engines, provisional. ChatGPT, Perplexity and Gemini returned an answer with an extracted business name on 100 of 100 questions, Claude on 76 under the analysis the predictions were scored on (64 under the current analysis). Aggregator and directory sites led ChatGPT's sources, birdeye on 43 and bestprosintown on 38.
- Scored
- Seven predictions scored on run 1, three missed. Every engine names a business on 90 or more, missed on Claude. Perplexity cites Yelp on 70 or more, missed at 7, with the instrument changed between prediction and run. ChatGPT's share of businesses' own sites between 35 and 55 percent, missed at 22, unclear until run 2. Met were the aggregator lead, a higher own site share on Gemini and Claude than on ChatGPT, run to run disagreement on every engine, and the home page indexed on day 0.
- Correction
- The record notes three reversals in one session, each where the writer stated something was absent from memory without checking.
- Status
- Indexing, search traffic and citation of the site itself are read through day 60.
Does one link from another site Google already crawls get a neglected page fetched?
A sentence with one link from stephanemorera.com to evoix.io/services/ai-seo, last crawled August 15. Live October 24.
- Predicted
- stephanemorera.com is recrawled within 7 days (high). The service page is crawled within 14 days (medium low).
- Result so far
- The linking page was recrawled in under 4 days. The service page was not crawled by day 14, 84 days after its last fetch, so that prediction missed. Google fetched it on day 19, the first time in 89 days, but Search Console does not list stephanemorera.com as a referring page, so nothing shows the link caused it.
- Status
- Verdict November 21.
Does rebuilding EVOIX's Google Business Profile with the right category, services and products change how people find it?
EVOIX Google Business Profile. Category changed to Marketing agency, six services and three products added, one post, October 24.
- Predicted
- Searches showing the profile at least double by day 30 (medium). A top search term contains "aeo", "ai" or "marketing" (medium). At least 2 customer interactions in 30 days (medium low).
- Result so far
- Day 14, November 7. The top terms had not changed, with one interaction since the edit. Google AI Mode named EVOIX first on two of five reads on October 24, a signed out Incognito read and a DeLand pull, and on none of 10 later runs. No change on any measured surface yet.
- Status
- Day 30 read November 23.
Does a first hand "how to get cited by ChatGPT" post reach Google's first page and the engines' own sources?
evoix.io/blog/how-to-get-cited-by-chatgpt, live October 22.
- Baseline
- ChatGPT's searches on the question all ran on OpenAI's own sites, and 35 of its 35 sources were help.openai.com.
- Result
- November 5. Indexed on day 0, met. 24 impressions, but no query row containing "chatgpt" within 14 days, not met so far. A recheck is due November 19 because Google hides low volume rows. Absent from the first page on 3 of 3, open until December 21. Cited by an engine on 0 of 12 runs, failed.
Does rebuilding a page Google declined, at the same address, get it into Google's index and into the AI engines' sources?
evoix.io/blog/aeo-vs-seo, declined after a May 9 crawl. Rebuilt in full and live October 22.
- Baseline
- Zero impressions in either 28 day window. Cited on 0 of 24 engine runs.
- Result
- November 5. Recrawled on day 0 and indexed by day 6, both met. 60 impressions in the first full window against 0 before, scored as met, though the prediction named searches containing "vs" or "geo" and the record does not split the 60 that way. Cited on 0 of 24 engine runs, failed. Google's first page by December 21 is still open. Google took the page back, and the engines still cited it on none of their runs.
Does an exact match domain enter the engines' sources for "Who is the king of AEO?" with no press release?
kingofaeo.io, one page, linked from his own sites. Live October 23.
- Predicted
- Indexed within 7 days (high). Claude cites it on at least 3 of 10 runs of the question (medium), ChatGPT 2 of 10 (medium low), Perplexity 1 of 10 (low). Google's top 10 on at least 1 of 3 pulls (low).
- Result
- Read November 7, a day late. Indexed, met. Cited on 0 of 30 runs on the three engines whose sources can be read. Absent from Google's top 20 on 6 of 6 pulls. The clause written to prove the idea wrong fired.
- Unpredicted
- Claude cited the domain on 8 of 9 runs of the contenders, Florida and Central Florida questions, each time next to his own sites. A Claude citation on the head question at day 7 was a single run that did not survive ten.
Does shipping everything at once, a wire release, video, social posts and owned pages, move the head question "Who is the King of AEO?"
A stack around the King of AEO claim, planned October 2 with no controls by design.
- Predicted
- ChatGPT names him on at least 4 of 10 runs of the head question (medium), Perplexity 3 of 10 (medium low), Gemini 2 of 10 (low), Claude 1 of 10 (low). Contenders and Florida questions hold at 9 of 10 or better.
- Result so far
- The wire release was dropped on October 2, so the two wire lines are scored not applicable. No video has shipped, and its prediction stands. Interim reads on October 21 and 28, three runs per engine. The head question did not move, 1 of 12 named on October 28 against 4 of 12 a week earlier, both inside noise. The two Florida questions held at 24 of 24 and the contenders question at 11 of 12.
- Defect
- A number animation on stephanemorera.com left the tallies at zero in the raw page, and Google indexed wrong figures, "10 of 40" and "20 of 80" for a true 39 of 40 and 80 of 80. Fixed October 2.
- Status
- No final read yet.
Does listing EVOIX's own tracker on EVOIX's own tools list get a brand new product named by the engines?
evoix.io tools list. The entry for In the Sources is marked as ours. Live October 1.
- Baseline
- Ship night, ten runs each. The product named on 0 of 10 by ChatGPT and 0 of 10 by Perplexity.
- Predicted
- ChatGPT names the product on at least 1 of 10 runs within 14 days. Perplexity on at least 2 of 10.
- Status
- No read after ship day is recorded in the file.
Does a question shaped King of AEO page on a second domain he owns enter the engines' sources?
stephanemorera.com/king-of-aeo, live October 2.
- Predicted
- Indexed within 7 days. Perplexity cites stephanemorera.com on at least 3 of 10 contenders runs, ChatGPT 1 of 10. Google's AI Overview names him on 0 of 6.
- Result so far
- Day 19, three runs per engine, provisional. Indexed. The site cited on 0 of 9 readable runs, not met. Google's AI Overview named him on 2 of 3, against the prediction, on a surface the page did not target.
- Confound
- The design called for the page to run alone for two weeks. Other King of AEO pieces shipped the same day, so nothing can be credited to the page alone.
Can a new list post enter a search result held by Reddit threads and LinkedIn posts?
"The 12 Best Answer Engine Optimization Tools for Small Businesses in 2026, Priced", live September 30 as an 11 tool list, 12 tools since October 1.
- Predicted
- Indexed within 7 days. At least one impression on a "tools" search within 14 days. Google's first page for one of three phrases by November 29.
- Result so far
- Indexed by day 1. Impressions arrived, met, but from pages 6 to 8, average position 54.8 in the 28 days to October 18. ChatGPT and Perplexity cited it on 0 of 10 runs each on October 2. Three scheduled reads did not run.
- Confound
- EXP-008 edited the page on day 1.
Does a press release move "Who is the King of AEO?"
Designed on September 30 around a paid wire release that was never bought. It ran on November 16 with a free PRLog release instead.
- Baseline
- November 16, ten runs per engine. ChatGPT named him on 2, Perplexity, Claude and Gemini on 0. No owned page cited on any of the 40. Google's AI Overview named him on 0 of 3.
- Result so far
- About an hour and 40 minutes after publication, the release was indexed and sat fourth in Google for "stephane morera". That search is his name. The head question is read at day 14.
- Status
- Day 14 read November 30.
Does merging and redirecting blog posts Google crawled but declined help the pages they point to, and does it cost AI citations?
evoix.io blog. Five redirects, three posts merged into one article, live September 30.
- Result
- Day 28, October 28. Google refetched 0 of the 5 redirected addresses, failed. The destination pages' share of site impressions went from 0.83 to 2.30 percent, 2.8 times against a predicted 1.2, met, a derived figure. The two control pages stayed out of the index, met. Google recrawled 26 of the 40 indexed pages in the 28 days against a predicted 30 or more, a miss the file expected. The AI citation guard was not scored, and the day 14 read was never run.
- Reading
- Merging content into an indexed page lifted that page. Redirecting declined pages did nothing visible in 28 days because Google never went back to them. The gain is credited to the merge, as a hypothesis.
- Correction
- The baseline was first read off a truncated table. The file corrected it, which made the prediction harder to meet.
A page Google had not crawled in four months might have been skipped as redundant. Does folding it into an indexed hub get the hub crawled, without losing AI citations?
evoix.io/services/ai-seo. Content merged, two redirects, 30 internal links rewritten, September 15.
- Result
- The hub was predicted to be recrawled within 7 days. Its last crawl was still August 15 at day 7 and still August 15 on October 22, 37 days after the fold. Google next fetched it on November 12, day 58 (EXP-014). ChatGPT kept naming EVOIX first on 3 of 3 runs and cited the hub on 0 of 3. Claude kept EVOIX as its top pick on 3 of 3.
- Reading
- Naming held and the hub stayed invisible to Google and ChatGPT. The redundancy explanation was reversed. EXP-014 later tested a different route to the same page.
Can an unknown name get onto the AI engines' King of AEO contender list with one owned page and a few social posts?
evoix.io King of AEO post, live September 12 at 10:45 PM Eastern (the post is dated September 13), plus his social accounts. No domain network, no paid placement.
- Baseline
- 0 of 9 engine runs and 0 of 3 Google pulls named him.
- Result
- September 30, ten runs per engine. Named on 39 of 40 contenders runs and 80 of 80 Florida and Central Florida runs. On the bare question "Who is the king of AEO?", 0 of 12, and Google's AI Overview answered James Dooley. A Florida AI Overview naming him was seen on two signed in browser reads, never on a cold pull, and stays provisional.
- Where it was wrong
- He predicted ChatGPT and Perplexity would not name him at day 7, and both did on 3 of 3. He predicted Claude 0 of 10 at the verdict and got 9 of 10. ChatGPT named him on none of the three questions asked three days after day 7, one run each, then came back to 10 of 10. Why Claude changed is not established.
- Raw runs
- king-of-aeo-measurement on GitHub.
Does ChatGPT pick regional "best agency" candidates by matching the region's name in page titles?
evoix.io. "Central Florida" added to the title and main heading of three pages, September 12.
- Predicted
- With no location signal in the request, ChatGPT names EVOIX on at least 3 of 10 runs by September 26, against 0 of 2 before.
- Result
- Read September 30, four days late. 10 of 10, beat. The second check, runs from a DeLand location, was never run.
- After
- Claude named EVOIX on 0 of 15 runs of the question from October 21 to November 5. Perplexity named it 3 of 3 on October 28, then 0 of 6 on November 5. The cause is not established. The file also carries two correction notes from October 22, one for comparing two different queries and one for four miscounts.