Brand Visibility in Chinese AI Engines: Doubao and Qwen

If your buyers are in China, they are not asking ChatGPT about you. They are asking Doubao. ByteDance’s assistant reached 345 million monthly active users in March 2026 according to QuestMobile’s Q1 report, which makes it the second-largest AI app in the world by monthly users, behind ChatGPT and ahead of Gemini.

Almost no Western AI visibility tool measures it. That gap is the subject of this article.

The short version

1. Doubao had 345 million monthly active users in March 2026, second globally by app users.
2. Alibaba’s Qwen passed 200 million monthly users and added roughly 126 million in a single quarter.
3. Google is blocked in mainland China, so the keyword tools most Western teams plan with cannot see this demand at all.
4. Of the tools in our comparison, one covers Doubao and Qwen, and it is ours. That is a statement about the category, not a boast: nobody has built for this market.
5. The opportunity is in English. The buyer is a Western company selling into China, and search demand for that sits at difficulty 0 with CPCs above $30.

The scale, with sources

For scale on the growth: Reuters put Doubao at 78.6 million monthly users in January 2025. Fourteen months later the figure is 345 million.

A caveat we would want if we were reading this: these numbers were read on 6 August 2026 and come from QuestMobile and AICPB via reporting, not from the companies. App-level monthly users is also a generous metric, and it is not directly comparable to how Western tools count assistant usage. Treat the order of magnitude as solid and the precision as approximate.

Why your keyword tool cannot see this

This is the part that quietly breaks Western planning.

Google is blocked in mainland China. Every keyword tool that Western marketing teams use, ours included when it comes to search volume, derives demand from Google. Asking a Google-based tool about Chinese demand is like measuring Madrid traffic with a sensor in Lisbon: the instrument is not wrong, it is somewhere else.

The practical consequence is that a Western brand can look at its dashboards, see no Chinese signal at all, and conclude there is nothing there. The absence is in the instrument.

This is why the honest version of this article does not include a table of Chinese search volumes. We cannot verify them with the tools we have, so we are not going to publish them.

What is different about optimizing for Chinese engines

Four differences that matter more than any tactic.

1. Different index, different sources. Doubao and Qwen do not retrieve from the same web your Western pages sit in. Baidu, WeChat articles, Zhihu and Xiaohongshu carry weight that Reddit and Wikipedia carry in English. A page that is authoritative in English may simply not be in the candidate pool.

2. Language is not optional here. The pull toward same-language sources is documented: in Temso’s analysis of over 7 million citations, French prompts were answered mostly with French sources, at 82.1% in AI Overviews and 72.3% in ChatGPT. There is no equivalent published study for Chinese that we could find, but the mechanism is the same and the language gap is far wider.

3. Hosting and access affect retrieval. A site that loads slowly or fails from inside mainland China is a site that is harder to retrieve there, regardless of how good its content is.

4. You cannot verify it from a browser in Europe. Checking what Doubao says about your brand requires querying it in the right market. This is exactly the measurement problem, and it is why the answer here is a tool rather than a manual check.

Where the commercial demand actually is

The interesting finding for a content plan is that the demand is in English, not Chinese. The person searching is a Western marketer trying to reach China.

Small volumes, zero difficulty, and CPCs that tell you what a lead is worth. Meanwhile the informational terms are enormous: doubao at 22,200 monthly searches with difficulty 18, and qwen at 74,000 with difficulty 37, both in the United States. People outside China are looking these engines up in numbers, and nobody is connecting that curiosity to the brand visibility question.

What to do about it

1. Find out whether you appear at all. Not a guess, a measurement, run from the right market with the language set explicitly. If your buyers are in China and you have never checked Doubao, you have no baseline.
2. Publish in Chinese if the market is real. A machine translation with an English canonical is not a Chinese page. If China is not commercially real for you yet, skip this: it is a serious commitment.
3. Check that your site works from there. Load time and reachability are retrieval prerequisites, not polish.
4. Build presence on the platforms that carry weight locally, which is the same earned-mention work as anywhere, aimed at a different set of publications.
5. Measure repeatedly, not once. Answers vary between identical runs. We asked one Spanish AI Overview question three times and got 11 cited domains, 4 of which appeared exactly once.

Where EchoWi fits, and where it does not

We track 12 engines including Doubao and Qwen, which as far as we can tell makes us the only tool in our comparison of fifteen that covers them. Evertune covers DeepSeek and stops there. Everyone else covers the Western set.

We are the wrong choice if:

• China is not commercially real for you. Paying for coverage you will not read is worse than paying nothing. Promptmonitor covers six Western engines at $29 a month.
• You need a full China marketing operation. Measurement is one piece. Local hosting, WeChat presence and Chinese-language content are a different, larger programme, and a specialist agency will serve you better than a dashboard.
• You want a guarantee. No study has established what moves visibility inside Chinese models specifically. We can tell you what they say about you. We cannot promise what will change it.

We are worth a look if you already sell into China or Southeast Asia, and you currently have no idea what 345 million people’s default assistant says when they ask about your category.

Common Questions About Chinese AI Engines

What is Doubao?

Doubao is ByteDance’s AI assistant, the same company that owns TikTok. It reached 345 million monthly active users in March 2026, which made it the second-largest AI app in the world by monthly users, behind ChatGPT.

What is Qwen?

Qwen, also written Tongyi Qianwen, is Alibaba’s AI assistant and model family. Its app passed 200 million monthly active users in early 2026 and added roughly 126 million in a single quarter.

Can I check what Chinese AI engines say about my brand?

Yes, but not reliably from a browser in Europe or the United States. The engine needs to be queried from the right market with the language set explicitly, and repeatedly, because answers vary between identical runs. That is a measurement problem rather than a browsing one.

Do I need Chinese-language content to appear in Doubao or Qwen?

Almost certainly, if you want to appear for Chinese-language questions. Retrieval pulls strongly toward sources in the language of the query, which is documented across engines and languages. There is no published Chinese-specific study we could find, so treat this as a strong expectation rather than a measured fact.

Why do Western SEO tools show no Chinese search data?

Because Google is blocked in mainland China and Western keyword tools derive volume from Google. The absence of a signal in your dashboard is a property of the instrument, not evidence about the market.

Which AI visibility tools track Chinese engines?

Very few. Of the tools in our comparison, ours covers Doubao and Qwen, and Evertune covers DeepSeek. The rest cover the Western set only.

Is Baidu still relevant for AI visibility?

Baidu runs Ernie and remains a major search index in China, so it matters for retrieval even where its assistant’s user numbers trail Doubao and Qwen. Reuters put Ernie at 13 million monthly users in January 2025, well behind the leaders.

Where this leaves you

The gap here is not subtle. The second-largest AI assistant on earth is one almost no Western brand has ever queried about itself, and the tooling reflects that: the category was built for English-speaking markets and has stayed there.

If China is not a real market for you, this is a curiosity and you should spend the money elsewhere. If it is, the first step is not a strategy. It is finding out what 345 million people’s default assistant currently says when someone asks about your category, and then deciding whether you can live with the answer.

Next: what GEO is and what the research supports, and the tool comparison with verified prices.