What GEO / AEO is (one minute)
In one sentence: GEO (generative engine optimisation, also called AEO) is making AI able to fetch you, understand you, and choose to cite you. You optimise the answer an AI gives, not a position in a list of links.
| Dimension | SEO (search) | GEO / AEO (AI answers) |
|---|---|---|
| What is optimised | Ranking in results | Mentions and citations in the answer |
| What users see | Ten blue links | A synthesised answer, often without sources |
| Your position | Page one or page two | Mentioned or non-existent |
| Key work | Keywords, backlinks, authority | Citable facts, entity consistency, source-pool presence |
| Measured by | Clicks, rankings | Mention rate, cited sources, source-pool hits |
Why now
- The entry point is moving. People ask AI "who can do this" instead of scanning results. If AI does not mention you, you do not exist.
- A good website is not enough. We measured our own: technical readiness 100/100, yet 0 mentions in 56 real AI answers (8 questions × 7 rounds). The bottleneck is the source pool, not the page. See the data (Chinese) →
- Information vacuum is opportunity. Ask AI about the Chinese customer-service standard and it names the wrong standard title. Whoever publishes correct facts first occupies the answer.
Likely effect and timeline
- Leading indicator (usually weeks 2–4): source-pool hit rate rises — you start appearing in the sources AI can retrieve.
- Outcome indicator (usually weeks 4–8): brand mention rate rises, your pages appear as cited sources, wrong information is replaced by correct facts.
- Honest boundary: AI answers are probabilistic and shift with models and platform policies. We promise a verifiable process, not a guaranteed citation.
Then the play is different from a national brand: do not chase national keywords — your customers are not there. Chase local phrasing instead: "who does enterprise knowledge bases in 〔your city〕", "〔city〕 best AI customer support".
AI decides whether a business really exists in a city from local sources. In our own measurement of local queries (Chongqing, 10 questions), the pool was dominated by local business directories and B2B listings (e.g. 11467, b2b168), job boards (BOSS Zhipin and similar) and Toutiao accounts, plus Douyin, 360 and CSDN content slots. National players ignore these, so local slots are easier to win — local GEO often moves faster than national.
GEO / AEO services: check · monitor · placement
The self-check measures your technical readiness. Whether AI actually mentions you depends on whether it can find you in the source pool — that needs real data. Three stages, each available on its own:
| Service | What you get | Cycle |
|---|---|---|
| ① GEO audit (starter) |
A report: whether AI can find you, which step is the bottleneck, which platforms to occupy first — from real query sampling, not a scorecard. | 5 working days |
| ② Monthly monitoring | A monthly report: brand mention rate, competitor comparison, source-pool movement, with verbatim AI answers so you can verify. | Monthly |
| ③ Content placement | Content placed on the platforms AI actually crawls, with a placement log; a crawlable landing page where needed. | Monthly |
Scope and fees are quoted per engagement, based on your industry, target questions and competitive intensity — we do not publish a generic price list. The flow is: free baseline audit first → see the findings → then decide what to do and what it costs. If you are interested, please use the contact details at the top of this page.
Why "being crawled" is not enough
AI answers are not assembled from the whole web. They are assembled from a retrieval pool — the set of sources the engine already trusts and can reach. If you are not in that pool, a technically perfect page still will not be quoted.
We measure three things, in this order:
- Reachability — do the crawlers that matter (including China-based ones) actually get your pages, or is the edge blocking them?
- Pool presence — when a real user question is asked, which sources show up, and are you among them?
- Answer share — in a batch of real AI answers, how often are you mentioned, and in what role (recommended / listed / ruled out)?
Start free: the self-check
Our online checker runs the technical layers against your domain: crawler access, structured data, entity consistency, sitemap and indexability signals, and it returns a weighted score with a prioritised fix list.
The interactive checker is currently Chinese-language; the audit logic and output structure are the same ones exposed in our GEO MCP.
Prefer to automate? Connect the GEO MCP instead: GEO MCP endpoints and tools.
What a paid engagement includes
- Baseline audit with per-item evidence, so every claim is checkable rather than asserted.
- Fix plan by leverage: what to do first, why, how to implement, how to verify.
- Atomic answers: one page per question a customer actually asks, written to be quotable.
- Monthly / quarterly measurement with the original AI answers attached, so the trend is verifiable.
We also publish our own measurement of our own site, including the parts where we score badly.
Honest boundaries
- A score is an observation, not a promise of inclusion or citation.
- Platform behaviour changes; results carry timestamps and must be re-measured.
- We do not treat
llms.txtas a ranking mechanism — it is checked for existence only. - Where a number cannot be obtained, we leave it blank and state why.
Want a written GEO report for your site?
Website AI visibility — audit, fixes, recurring measurement
Start from the knowledge base