GeoOne rigorous pipeline. Eight modules.
Every GeoAIVO audit runs the same transparent process — from modelling who your buyers are, to querying real AI engines, to a weighted score and an action plan.
How the score is built.
A fixed-weight rubric combines four dimensions into a single 0–100 AIVO Score, so it's comparable month over month and across brands.
Inside the 8 modules.
USER_PROFILE — persona & question mining
We use LLMs to model your buyer personas and usage scenarios, then extract the high-intent questions they actually ask AI — split into two buckets: directly related ("best imported serum for X") and generalized related ("how to fix sensitive skin"). These questions become the test set everything else runs on.
Method: LLM persona analysis + search-intent miningINFRA_EVAL — infrastructure evaluation
We assess the crawlable base the engines trust: your official site, owned media (WeChat, official accounts) and authority-media footprint. Thin or inconsistent infrastructure caps how confidently AI can cite you.
Method: web crawl + LLM summary + media searchCOMPETITOR — competitive set & comparison
We discover the brands the engines actually recommend in your category and build a structured, engine-by-engine comparison table — the benchmark your Share of Model is measured against.
Method: competitor search + LLM structured comparisonAI_SEARCH — real visibility testing Core
The heart of the platform. We put the mined questions to multiple Chinese LLMs — DeepSeek, Doubao, Kimi and more — with real prompts, and record whether your brand is mentioned, how, and in what position. Not estimates. Actual answers.
Method: live multi-engine querying + mention captureGEO_EFFECT — aggregation & comparison
We run statistics over the AI_SEARCH results and visualize them: per-engine Share of Model, overall SoM, and how you compare platform by platform.
Method: statistical aggregation + visualizationSENTIMENT — narrative health
When engines do mention you, how do they frame you? Using negative-keyword lists and LLM sentiment analysis, we classify your narrative — leader vs. reliable vs. budget — and surface any risk descriptors.
Method: negative-keyword lists + LLM sentimentAIVO_SCORE — weighted scoring
The four dimensions — Visibility, Infrastructure, Competitiveness and Sentiment (25% each) — combine into a single, comparable AIVO Score out of 100.
Method: fixed-weight rules-based scoringSUGGESTION — priority actions & roadmap
Finally, we turn scores and gaps into a prioritized, impact-ranked action list and a 12-month roadmap — the exact steps to move your Share of Model.
Method: rule + LLM roadmap generationRun the full pipeline on your brand.
See all 8 modules in action in the live demo dashboard.