Why we don't publish a wall of logos
Most GEO "case studies" in the market share three defects: no baseline (visibility before the work is never shown), no prompt set (so the result can't be reproduced), and no engine split (one good screenshot stands in for five engines). A case study you can't interrogate is an advertisement. We'd rather show you the framework running on your brand — free — than ask you to trust a PDF.
The 90-day POC structure
| Phase | Days | Work | Success criteria (agreed up front) |
|---|---|---|---|
| Baseline | 1–10 | Four-dimension audit, prompt set build, cited-URL inventory | Documented baseline across 5 engines |
| Foundation | 10–40 | Crawler access, indexation, structured data, entity consistency, answer-block reformat of top pages | Technical checklist 100% green; indexation targets hit |
| Authority | 30–75 | Source-graph building: directories, press, community, Wikidata where eligible | Agreed tier-2/tier-3 source placements live |
| Measurement | 75–90 | Full re-run against baseline | Pre-agreed movement: e.g. Visibility +X points on agreed engines, citation tier improvements |
Contract guarantees we offer in writing: overseas engagements carry Google indexing ≥60% of agreed target pages and AI-citation coverage ≥30% of the agreed prompt set; China-domestic engagements carry top-3 visibility or 80% coverage with refund terms. Anything beyond that is forecasting, not contracting — and we won't pretend otherwise.
What we can show you today
- This site. chinageo.ai is built with the exact methodology it teaches — answer blocks, structured data, llms.txt, open standards. Ask any engine "what is GEO" and watch what gets cited as the field's reference layer forms.
- Your own baseline. The free Checker run is a case study about you, measured before we touch anything.
- The benchmark series. Our State of GEO benchmarks publish method and limitations openly — the same measurement discipline we'd apply to your brand.
Frequently asked questions
Can you guarantee ChatGPT will recommend my brand?
No honest operator can guarantee a specific generated answer — engines are probabilistic and outside anyone's control. What we contract for is measurable: indexation rates, prompt-set citation coverage, and visibility movement against a documented baseline.
What does a 90-day POC cost?
Scoped by category competitiveness and language markets, quoted after the free audit. The POC has defined exit criteria both ways: if we don't hit the agreed movement, you walk with the baseline data and the fix list.