Sitegravity

How we built it

The system that produces your numbers is built by us. Here is what it does, and where it stops.

We built the measurement system ourselves

It is not a reseller's dashboard and not a spreadsheet of screenshots. The founder built it, and every number in a report can be traced back to the individual answers.

What it does

  • Question bank: Questions written by us from real customer situations, one set per language, without brand names.
  • Several engines, several languages: ChatGPT, Gemini, Perplexity, DeepSeek, Doubao and Yuanbao, in English, Simplified and Traditional Chinese and Malay (Doubao and Yuanbao in Chinese only).
  • Repeated sampling: Every question is asked several times on every engine, and we report the average with its range.
  • Business extraction and merging: An AI model lists every business named in an answer; we merge spelling variants, Chinese variants and company suffixes into one business.
  • Source classification: Every website an answer cites is labelled your own site, a competitor's site, a third-party site (directory, media, review, tourism board) or "unknown" when we have not confirmed whose it is.
  • The same questions every month: Tracking re-runs the same set, so a change in the numbers is a change in AI, not in the questions.

What it cannot do

  • We test through the engines' APIs, not through the consumer apps, so results can differ from what a person sees in the app.
  • Doubao is currently measured without web search, so it shows what the model learned in training.
  • DeepSeek and Yuanbao use third-party search (Exa), not their own.
  • We measure whether AI mentions and recommends you, not whether customers then buy.

Read the full method and limits →

See where AI stands on your brand

Get a free AI visibility check