Current edition
Q4 2026 · Current
AI-visibility benchmarks: 6 industries × 4 engines
Thirty buyer prompts per industry, three samples per prompt, per-engine visibility and citation-tier breakdowns. Methods and limitations published.
Read the benchmarks →Foundations
GEO & AI search statistics 2026
The sourced, dated industry numbers: AI Overviews reach, ChatGPT users, CTR impact, zero-click rates — with methodology notes.
See the statistics →Method
How we measure
Prompt-set design, three-run sampling, per-engine scoring and aggregation weights — reproducible by anyone.
Read the methodology →Editorial policy for data
- Methods before numbers. Every edition publishes its prompt sets' design rules and sampling protocol.
- Limitations stated. Panel sources, sample sizes and known biases are declared on the page, not in a footnote PDF.
- Ranges over points. Where independent measurements disagree (and they do), we publish the range and say why.
Frequently asked questions
How often is State of GEO updated?
Quarterly for the benchmark series, with the statistics page refreshed as major sources publish new figures. Subscribers to the monitoring service get monthly movements between editions.
Can I cite your benchmarks?
Yes — that is what they are for. Cite 'ChinaGEO State of GEO, Q4 2026' with the URL. The methodology sections exist precisely so your citation survives scrutiny.