GEO is a monthly service that improves how your brand appears in AI answers in China. We start from the questions your buyers ask, find missing or inconsistent evidence, improve the content, and check again every month.
Doubao
Qwen
DeepSeek
Kimi
Tencent Yuanbao
Baidu ERNIE
Before a buyer contacts you, they may read a news story, ask an AI engine to compare options and search Xiaohongshu. In a GEO check, we record whether agreed buyer questions return your brand, current facts and visible citations, with the question, engine and date.
GEO shows you what those answers say today, then works on the sources behind them: product explanations, FAQs, comparisons and the evidence that supports your claims.
Pick a question and an engine to see what we record: whether you’re mentioned, which source the engine cites, whether the facts are right, and what we’d work on first.
1 · Pick a buyer question
2 · Pick an engine
Illustrative example with fictional brands. Real checks record the engine, the question as asked, the date and the exact wording, with screenshots.
In practice
Three categories, three engines: software, pet food and cold-chain logistics. Click any screen to see it larger.
Engines we check
The free GEO check covers Doubao, DeepSeek and Qwen. Every monthly plan covers all six.
Developed by ByteDance
All monthly plans
Free GEO check
Developed by DeepSeek (Hangzhou)
All monthly plans
Free GEO check
Developed by Alibaba
All monthly plans
Free GEO check
Developed by Moonshot AI
All monthly plans
Developed by Tencent
All monthly plans
Developed by Baidu
All monthly plans
It works for B2B and consumer brands alike. The right starting point depends on what your buyers ask and how they decide.
Tick what sounds like you
Every month follows the same loop: research, check, improve, check again. Here is the work inside it.
Question list
Example formatYou always know what stage we’re at and what it should produce.
Steps 01–02
You share your category, audience, products, market position and the buying decision you want to influence. We agree the questions and engines.
You provide product facts, existing content and evidence. We agree who approves facts, publishing and technical work.
Steps 03–04
We check the agreed questions and record mentions, citations and accuracy, with wording, engine and date.
We turn findings into tasks, such as a clearer product explanation, missing evidence or inconsistent descriptions.
Steps 05–06
We write and revise the agreed content. You check the facts and approve it.
The responsible team puts the approved work live. We log every change.
Steps 07–08
We repeat the agreed checks and compare them with the baseline, recording mentions, citations and accuracy separately.
Together we review completed work, remaining gaps and new priorities.
Then the monthly cycle repeats: research, check, improve, check again.
Every plan covers all six engines and includes brand analysis and a consistent AI positioning, core content optimisation, article indexing and tracking, AI recommendation monitoring and knowledge base upkeep.
Plans differ in how many buyer questions and keywords we cover. The minimum term is 3 months, and longer terms cost less per month.
Basic plan
From US$1,008/month
Based on a 12-month contract
Contract length
Minimum term 3 months. Longer terms cost less per month.
Builds your GEO content base around core use cases.
What this plan covers
Buyer questions covered
All six engines in every plan: Doubao, Qwen, DeepSeek, Kimi, Tencent Yuanbao and Baidu ERNIE.
Included in every plan
Monthly prices are based on a 12-month contract. External media buying and website development are scoped and priced separately. KPIs, such as a target mention rate for agreed questions, can be written into the contract with how they are measured and what happens if they’re missed.
Knowing who owns each part keeps the work moving and the results honest.
We describe what we observe, with the question, engine and date, so you can check it yourself.
June: the brand is not in the answer.
September: named, with its Chinese launch and WeCom support cited.
How we measure
Example format[Brand] · GEO monthly report · [Month]
| Questions checked | [n] agreed questions × [engines]Asked in Chinese, with date |
|---|---|
| Mentions | [x] of [n] answers (baseline [y])Exact answer wording kept |
| Visible citations | [sources the answers named]Recorded separately |
| Factual accuracy | [errors found] · [errors now fixed]Checked against your approved facts |
| Changes made | [pages and FAQs updated]Change log with dates |
For the agreed questions we record brand mentions, visible citations and factual accuracy separately, then compare each month with the baseline.
What we don’t promise
Outside KPIs written into your contract. AI engines choose their own answers.
Can’t find your answer? Ask for a free GEO check and we’ll walk you through it.
Tell us three things about your brand. We’ll ask AI the questions your buyers ask and send you a one-page summary.
5 buyer questions × 3 engines: Doubao, DeepSeek and Qwen
A one-line conclusion, a results table and the 3 main issues, with screenshots
Sent within 3 working days of receiving your details
An optional 15-minute walkthrough, with no obligation
The one-page summary you’ll receive (template shown). It reports only what we observed, with engine, question and date.
Other services
Buyers move between Chinese media, AI answers, social accounts and creators. Add a second service where the next gap is.
