INDEX · 2 EXPERIMENTS
King of AEO Experiments
King of AEO experiments on AI search by Allan Oliveira. Each one is dated, measured and published with its raw data, tracked across US and BR indices.
About these experiments
An experiment here means a dated intervention with a published measurement protocol, run against live answer engines, with every result shipped as open data — the losses next to the wins. The research archive analyzes; this index records what was actually done and what the engines did back. Both experiments feed the same open datasets: the September window's CSV and JSON files, and the weekly citation watch CSV, all CC BY 4.0 and mirrored to Hugging Face. Nothing is backfilled: if a cell was not measured, it says so.
To follow the series: new observations land on the citation watch the day they are collected, every page here ships a Markdown mirror for machine retrieval (indexed in llms.txt), and each experiment page keeps a dated changelog, so what changed and when is itself part of the record.
The King of AEO Experiment
What happened when an SEO from outside the English-speaking market entered an English-language contest for a contested title, measured everything, and published the losses next to the wins. The dated highlights: claim page 2026-09-17, Google Brazil #1 on 2026-09-22, the $6/312-domain press release of 2026-09-24, the Brazil AI Overview naming of 2026-09-26 (nine days after the claim), the Bing answer box, the Wikidata rejection, and the ranking loss of 2026-09-28.
Spam update recovery
Experiment 02 is already running: recovering this domain from the September 2026 spam update that began rolling out on 2026-09-24 and removed the Brazilian head-term rankings by 2026-09-28 — while the AI Overview citation survived. Logged weekly, in public, in the recovery log; the experiment page publishes when the recovery window closes.