GEO Goes Corporate: China's Tech Giants Move In on AI Search Visibility

For two years, getting a brand into Chinese AI answers was a gray trade. Small agencies sold "source placement" packages for ¥50,000 and up, promising mentions inside AI responses. There was no standard, no shared measurement, and no way to verify what you bought.

That changed in the first week of September 2026. Tencent, Alibaba, 360 and Baidu each put their name behind a GEO product within days of each other. GEO — generative engine optimization — stopped being a craft and started being a product category.

For overseas brands, the timing matters. The market just validated AI visibility as a budget line. But the tools arriving are built for Chinese domestic small businesses, and the gap between "measured" and "visible" is about to define who wins.

🗞 What happened this week

Four moves, four different strategies:

• Tencent launched AnswerBit, an AI search brand visibility platform. It monitors Doubao, Yuanbao, DeepSeek, Kimi and Qwen, and records whether your brand is mentioned, at what position, whose content gets cited, and whether the framing is positive or negative. Tencent built it as a measurement layer rather than a placement service.
• Alibaba Cloud added GEO/SEO diagnostics to Wanxiaozhi 3.0. Enter a URL and it reports brand mention rate, site citation rate and average ranking, split by platform (Doubao, DeepSeek). A single "AI optimize for me" button then rewrites code structure and content tags. The pitch is explicit: you don't need to understand SEO.
• 360 released Zhijian GEO 2.0, a full chain from brand diagnosis to knowledge graph construction to AI content generation to data traceability. Its positioning — "make your brand the model's trusted answer" — is the clearest statement yet that GEO is now about trust, not volume. BlueFocus's agency arm has already signed on as a service partner.
• Baidu has a GEO solution pricing sheet circulating in the market: roughly ¥5,000 per quarter, ¥10,000+ per half year, ¥14,000+ per year. Whether this is an official Baidu product is still disputed — but the Baidu name is attached, and the price point targets small and mid-sized businesses.

The scale behind these moves explains the rush. MIIT's CAICT puts China's GEO services market at ¥28.6 billion in 2026, up 125% year on year. Gartner estimates traditional search traffic is falling about 25% annually while AI search traffic grows about 125%. QuestMobile counts 820 million monthly active AI search users in China. And 37% of B2B buyers now use AI search to shortlist suppliers — which means a brand invisible in AI answers loses roughly a third of its consideration set before a sales conversation even starts.

🔄 Why now: from black-box craft to rules business

Every giant is doing the same structural thing: taking its own content reservoir and turning it into the supply side of AI answers.

ByteDance feeds Doubao with Douyin's short video and image-text corpus. Baidu feeds ERNIE with two decades of web search crawl. Alibaba feeds Qwen with structured product data from Taobao, Tmall and 1688 merchant backends. Tencent feeds Yuanbao with WeChat Official Accounts, Channels and Tencent News. Each platform answers from what it already owns.

That is why the giants can productize GEO and a five-person agency cannot. The agency was selling information asymmetry — write articles, place them on media sites, seed sources, report rankings, charge a package fee. In 2026 an AI can write the article, monitor the keyword and rebuild the site. When the work becomes a button, the margin on knowing the trick collapses.

The analogy that fits: SEO didn't die, but the people who made money selling links largely did. First-generation GEO vendors are facing the same compression.

The deeper shift is that the interpretation of GEO rules is moving to the platforms. When Alibaba decides how product data enters an AI summary, or Tencent decides what AnswerBit counts as a "positive mention," that becomes the de facto standard. Brands no longer optimize against a public algorithm — they optimize against each platform's published product.

🧰 What the tools do — and where they stop

This is the part that matters for a brand headquartered outside China. The new tools are genuinely good at three things:

1. Measurement. You can finally see mention rate, citation share and sentiment per platform. That is a real step forward from "we sent you a screenshot."
2. On-site optimization. Alibaba's one-click rewrite of structure and tags will help any site that is technically invisible to crawlers.
3. Content generation. Producing Chinese-language content at scale is no longer the bottleneck.

But all four tools stop at the same wall, and the wall is where overseas brands live.

They measure, they don't place. AnswerBit tells you that Doubao cited a competitor. It cannot get your brand into the source Doubao prefers.

They optimize your property, not your standing in other people's property. AI answers in China are built from third-party authoritative sources — vertical leaders like Ctrip for travel, Autohome for autos, Sina and Toutiao for finance. We covered this in our September 6 analysis of the "fewer sources, deeper content" shift: platforms are citing fewer sources but reading the top ones far more deeply. A tag rewrite on your own site does nothing to put you inside Ctrip.

Every tool covers one ecosystem. Alibaba optimizes for Qwen's supply chain. Tencent monitors from Tencent's vantage point. Baidu's effort is tied to ERNIE and Baidu search. 360's is tied to 360. No vendor gives you consistent visibility across Doubao, DeepSeek, Qwen, Yuanbao, Kimi and ERNIE — which is exactly the set your Chinese buyers are spread across.

So the giants didn't remove the moat. They moved it — from knowing the trick, to owning the asset: verifiable Chinese content and real standing in the sources each platform trusts.

⚖ Compliance: three documents reset the bar

While the platforms were productizing, three governance documents landed from three independent directions:

• T/CAPT 026—2026, the GEO Trusted Information Dissemination and Information Ecosystem Governance Specification, published in August and led by Xinhua and the Xinhua National Key Laboratory. It sets operating standards across the full chain: intake, content review, corpus access, optimization execution, distribution control, monitoring and traceability, risk handling.
• The GEO Industry Self-Discipline Convention, signed in Beijing on March 14 under the China Publishing Promotion Association. Eight chapters, 33 articles, with a black-hat GEO negative list and a joint disciplinary mechanism.
• The GEO Red Paper 2026, published by National Business Daily's AI think tank with 40+ institutions. It defines five risk dimensions: improper purpose, information distortion, technical manipulation, scope of impact, and reversibility difficulty.

Together these make compliance the dividing line. And it hits overseas brands hardest, for a structural reason: the most common GEO failure mode for a foreign brand is not manipulation, it is distortion — a claim that reads stronger in Chinese than in the English source, a qualification the company holds in the EU but not in China, an "award" from an organization no Chinese platform recognizes. Under a five-dimension risk test, that is exactly what gets a brand filtered out of an AI answer.

Fixing this needs someone who can read Chinese regulatory context and your English source material in the same pass. A one-click optimizer cannot do it.

🎯 What overseas brands should do now

Four moves, in order:

1. Buy measurement before placement. If a vendor cannot show per-platform mention rate, citation share and sentiment, you are still buying screenshots.
2. Identify the 2–3 gatekeeper sources in your category. Ask each platform your buyer's actual question and record which sources it cites. That list is your shelf space — not your own blog.
3. Make every claim verifiable in Chinese. Facts must match across your English site, your Chinese site and every third-party mention. Inconsistency is the fastest route to being dropped.
4. Plan for cross-platform coverage from the start. No single vendor and no single platform tool covers the full set. Budget for the portfolio, not the button.

🧭 The BPP position

The giants just made GEO legitimate, measurable and cheap — for companies that can operate inside the Chinese internet.

Overseas brands usually cannot, for three reasons unchanged by any of this week's launches:

• Entity and access. Most GEO tooling, service partners and authoritative Chinese sources require a Chinese business entity, mainland settlement and local account management.
• Cross-platform consistency. Your visibility has to hold across six assistants that no single vendor covers.
• Compliance under the new rules. Distortion risk sits at the intersection of your English source copy and Chinese regulatory expectations.

BPP works at exactly that intersection: Chinese-entity access without a Chinese license, cross-platform AI visibility across the Chinese assistants, and compliance-reviewed Chinese content built to survive the three-document test.

The tools got better this week. The wall stayed where it was.