Qwen is the leading assistant in Chinese-speaking markets. If you sell there, buy there or compete with brands from there, it is the only way to know what is being said about you in that language.
Free plan, no card required · Nine AI assistants
Out of every 100 answers, how many name your brand
Last 30 days, 10 questions asked every day
Of the nine engines we measure, Qwen is the most specific. It makes no sense for most local businesses and complete sense for anyone exporting, importing or working with Chinese manufacturers and distributors.
The reason is simple: an answer generated in Chinese about your category is built from Chinese sources, and those are not the sources ChatGPT or Gemini use. You can have an excellent reputation in the West and a wrong, outdated or entirely absent description in that other set of sources.
If your name is in that answer, you get the customer. If it is not, you never find out: there is no list of results to check your position in and usually no click either, so nothing shows up in your website statistics.
For the Western engines there are academic studies that let us talk in numbers. For Qwen far less has been published, so we would rather not invent precision we do not have.
What we do know is that the underlying patterns repeat in studies of Chinese-language generative engines: the already-known brand comes out reinforced, only a small share of the sources the engine retrieves actually make it into the answer, and recent content weighs more on time-sensitive questions. Measuring your own case remains the only way to know where you stand.
We said earlier that little has been published about Qwen specifically. About Chinese-language generative engines in general there is research, and it is harsh: a study of 614 queries and 160,860 citation records found that only 8.3% of the brands the engine retrieves end up appearing in the answer, and that barely 12.4% of sources carrying contact details get those details shown.
There is another finding that changes how the work has to be set up: the language of the question decides the sources. Another study measured that ChatGPT cites more when asked in Chinese than in English, and that Google does exactly the opposite. Asking in English to find out how you are doing in the Chinese market is useless: you have to ask in Chinese.
Out of every 100 brands the engine reads, how many end up in the answer
Chinese-language generative engines, 614 queries and 160,860 citation records
Source: T. Zhen, Y. Liu, G. Zhang and others, arXiv:2607.15771. Each bar is on its own total.
We ask your 10 questions to Qwen continuously and store every answer whole: whether it names you, where, in what tone, next to which competitors, which pages it read to build the answer and what it searched for first. You read the month, not the day.
The full method, with the research behind why it is measured this way and what can actually be changed, is in how we measure whether an AI names your brand. And there is nothing to install on your site.
Because these engines do not answer the same thing twice: only three to five out of every ten pages they use repeat from one query to the next. A single answer tells you almost nothing; how many days out of thirty you came up does. It is set out with the data in how we measure.
Yes. The free plan asks your questions to ChatGPT every day at no cost and with no card, so you can see how it works. The other engines come with the paid plans.
In whichever language you write your questions. If your target is the Chinese market, the sensible move is to write the prompts in Chinese, because the sources the engine retrieves change completely with the language of the question.
From the Pro plan, alongside AI Mode, DeepSeek and Copilot.
We ask your questions to Qwen and the other assistants and keep the history, so you can see how many days you came up and who came up instead of you. Free plan, no card needed. Full detail on the pricing page.
Start tracking freeAlso check your brand in ChatGPT · Google AI Mode · Google AI Overviews · Perplexity · Gemini · Copilot · Claude · DeepSeek.
