How the snapshots are made
Every public snapshot uses the same fixed procedure. This page says how we ask, how we count, and where the data falls short, so you can check it.
AI Visibility Check › How the snapshots are made
How we ask
Each question is asked in a new conversation by a person, and the original answer is saved. For every answer we record the date, the model and version, whether web search was on, and the question language. Chinese-language models are asked in Chinese, overseas models in English. We use our everyday signed-in accounts, so an account's personalization settings can affect an answer; where that is the case, it is noted under "How we asked".
How the questions are chosen
Each category uses 10 questions. Six do not name any brand and show whether a brand comes up in general category questions. Four name the brand and only record whether the brand appears in the answer and which other brands appear with it; we do not publish opinions about the brand.
How "mentioned" and position are counted
A brand counts as mentioned when its name appears in the answer. Position is the order in which brands first appear in that answer: first, among the first three, or later. For a model, we take the earliest position across its questions. A brand's name is counted as written; we only treat the same brand's English and Chinese names as one.
Why brands are listed alphabetically
The snapshots record what appeared, not a ranking. Alphabetical order keeps the tables from being read as a league table.
Where the data falls short
Answers vary by person, time, region, language and model version. Ten questions is a small sample, and AI models can be wrong. That is why every page shows its date and conditions, and why the snapshots are refreshed.
Corrections and removal
If you think a page is wrong, or you want pages about your brand removed, write to hello@sharehub.fun. We will check and reply within 3 working days.
How long original answers are kept
We keep the original answers to check the data and to handle correction requests, for the period set out in our terms of service.
Want the full picture?
This snapshot uses only 10 questions. A full report covers more questions and models:
- 4 to 10 models, including China-based models such as Doubao, Qwen and DeepSeek
- 30 to 50 questions covering recommendation, competitor comparison and reputation
- A full competitor benchmark and analysis of the sources cited
- Monthly re-checks: the same questions re-run each month and compared with the previous month
- 3 recommended fixes for your website and third-party content; the deeper plan includes a re-check after 30 days
Why check regularly: more people now ask an AI model before they buy, and the answers change from month to month. A brand that is recommended today may not be next month, and one that is missing today may appear after its website and third-party coverage improve. Re-running the same questions on a schedule is how you tell which way it is moving.
From $99, delivered within 5 business days after payment is confirmed.
Want monthly tracking? Email hello@sharehub.fun with your brand and category and we will reply with options.