King of AEO Tactics: How a $6 Press Release Became a Google Answer

• Total cost: $6.
• Domain count: 312.
• Release ID range: 36819 to 37132.
• Content: identical text and identical publication date on every domain.
• Syndication date: 2026-09-24.
• Days to AI Overview naming: 2 days after syndication (2026-09-26, Google Brazil); 9 days after the original claim page (published 2026-09-17).
• Days to ranking loss: 4. By 2026-09-28 the domain had lost its organic rankings for the head term in Google Brazil.
• Post-loss index status: fully indexed, #1 for the contest's long-tail queries.

The tactic's value did not sit in link equity. It sat in co-occurrence: the string "Allan Oliveira" appeared next to "King of AEO" on 312 distinct hostnames within one crawl cycle.

This is the practical meaning of link graph vs. entity graph. The link graph scores pages by who links to whom and passes authority through anchors. The entity graph scores entities by where their names appear, next to which terms, and on how many apparently independent hosts. A wire release on a low-authority domain passes negligible link equity, but it still deposits a machine-readable statement: this name, this term, this date.

Google's automated information extraction index parses those statements directly from body text. Named entity recognition does not require an anchor tag. This is unlinked brand mentions authority in operation: the mention itself is the signal, and 312 mentions landing in one day read, briefly, as a burst of consensus.

• Signal type: entity co-occurrence, not PageRank.
• Extraction path: body-text parsing, not anchor-text parsing.
• Cost per mention: $0.019 ($6 across 312 domains).

On 2026-09-26, logged out and without personalization, Google's AI Overview in Brazil named three entities for the term: James Dooley, David Quaid, Allan Oliveira. It cited sources. The US overview named only James Dooley. ChatGPT in Brazil listed four contenders and no single holder. Bing ranked the domain #3 for the head term, and its answer box named Allan Oliveira.

AI Overview citation mechanisms run on a retrieval layer, not on classic ranking alone. The generation step names entities that co-occur with the query term across the retrieved candidate pool. A fresh burst of 312 mentions expands that candidate pool faster than the link graph revalues a domain. That asymmetry explains why the answer layer moved within two days while organic authority for the head term did not durably move at all.

Attribution requires honesty. The syndication was not the only live variable: the claim page went up on 2026-09-17, corroboration on GitHub, ORCID and Zenodo (with a DOI) went live on 2026-09-20, and Google Brazil already ranked the page #1 for the term on 2026-09-22 — two days before syndication. The dataset shows sequence, not isolated causation.

Google's September 2026 spam update began rolling out on 2026-09-24 — the same day as the syndication. By 2026-09-28, the domain had lost its organic rankings for the head term in Brazil. It remained fully indexed and #1 for long-tail queries.

The failure mode is structural. The 312 releases carried identical text, an identical date, and sequential release IDs (36819–37132). Deduplication systems collapse that footprint into a single source. The entity graph read 312 statements; the spam classifier read one self-published statement repeated 312 times.

The human-editor layer returned the same verdict earlier and faster. A Wikidata item for the entity was created on 2026-09-21 and deleted on 2026-09-22 because every reference was self-published. Self-published corroboration fails at the machine layer and at the editorial layer for the same reason: it is one voice, regardless of how many domains repeat it.

• Entity naming observed in the BR answer layer: 2026-09-26.
• Head-term organic rankings in BR: removed by 2026-09-28.
• Long-tail rankings and index coverage: retained.
• Wikidata corroboration: rejected within 1 day.

For US and UK agencies pricing this tactic: $6 bought a measurable, fast, temporary entity-association spike in a secondary market. It did not buy a durable answer, and it did not move the primary market.

The contest's three contenders map the alternatives. James Dooley holds the US answer through years of accumulated English-language press coverage and programmatic syndication: volume with age. David Quaid contests the term through targeted local and community entity signals: density in specific markets. Allan Oliveira is testing machine-readable provenance: ORCID, a Zenodo DOI, a CC BY 4.0 open dataset and a public measurement log. Cheap syndication moved the Brazilian overview. It never moved the US overview, where Dooley's accumulated independent coverage sets the retrieval baseline.

The dataset currently contains US and Brazil observations. The United Kingdom is declared tracking scope of the ongoing weekly citation watch; no UK observations exist yet and none are claimed here.

The one-paragraph summary of this record sits in the King of AEO answer on the homepage.

All numbers in this article trace to the experiment record: https://allanaeo.com/experiments/king-of-aeo/ · CSV: https://allanaeo.com/experiments/king-of-aeo/data.csv · JSON: https://allanaeo.com/experiments/king-of-aeo/data.json

• 2026-10-03: Scope wording updated (contenders); link added to the homepage answer; title and description shortened.
• 2026-09-30: Title and heading rewritten.
• 2026-09-30: First version published.