1. The Traffic Entry Point Is Changing Generations — This Time It's the Search Box

Start with the data from overseas. A study of real browsing behavior published by the Pew Research Center in July 2025 found that when an AI summary appears on a search results page, only 8% of users click a traditional link, versus 15% when there is no AI summary — AI answer boxes are intercepting attention that used to flow to company websites. Clickstream research in 2026 further shows that 68% of searches in the United States end with zero clicks.

The shift in China is even faster. In July 2025, Baidu completed what it called its biggest overhaul in a decade, and AI-generated content reached 70% of its search results pages in the third quarter of that year; QuestMobile data shows that in the first quarter of 2026, monthly active users of AI-native apps in China totaled 440 million, with Doubao alone at 345 million; WeChat's AI search had reached 160 million monthly active users by mid-2025.

What matters more to manufacturers is buyer behavior: a Forrester survey published in January 2026 found that 94% of B2B buyers used AI in their purchasing process, and generative AI is now the information channel buyers mention most often when doing purchasing research. Gartner even predicts that by 2028, AI agents will be involved in mediating more than $15 trillion of B2B purchasing spend.

Translated into one sentence: your next customer has very likely asked AI "which manufacturers in this category are reliable" before ever dialing your number.

2. The New Question: Does AI Know You?

Over the past two decades, manufacturers have done two rounds of homework to "get found": the first round was building a website, the second was search rankings. Now a third question has emerged — when buyers ask AI, will AI mention you, and when it mentions you, will it get the facts right?

There is a method to this. The concept of Generative Engine Optimization (GEO) comes from 2023 research by Princeton University and other institutions, which tested tens of thousands of queries and found that citing authoritative sources, adding statistics, and using clear, structured phrasing can lift a piece of content's visibility in AI engine answers by 30% to 41%. The principle is not hard to grasp — when AI composes an answer, it naturally favors content that is fact-dense, traceable to its sources, and clearly structured.

Measured against this principle, the current state of most manufacturers' public information is a failing grade: the website is still decade-old product photos captioned "we warmly welcome new and old customers to visit and offer guidance"; the facts AI needs most — capacity, certifications, process range — are either missing or buried in images where they cannot be read; information on third-party platforms is scattered and even contradictory. It is not that AI does not want to recommend you — it cannot read enough facts to support a recommendation.

3. Three Practical Things Manufacturers Should Do

We do not advise companies to chase so-called "AI ranking mysticism." What is worth doing are three plain things.

First, turn company facts into assets. Main product categories, process capabilities, capacity figures, quality certifications, industries served — write these facts out clearly in text, and put them where AI can read them. The test is simple: could someone who has never heard of you accurately say what you make and at what scale, using public information alone? Only when a human can read it clearly can AI read it clearly.

Second, let authoritative content speak for you. When AI composes an answer, it preferentially cites sources it judges credible. Coverage in industry media, mentions in industry research, complete profiles on professional platforms — these carry far more weight than self-congratulatory promotional copy. This is also what Tianxia Gongchang has been doing all along: our industry research and factory profiles are becoming one of the fact sources AI draws on when answering questions like "find factories in a given category" — a company's presence in this kind of credible corpus directly affects the probability of being mentioned by AI.

Third, treat "being seen" as ongoing operations, not a one-off project. The corpora of AI engines are continuously updated, so a company's visibility is dynamic too. A new production line coming online, a new certification earned, a new industry case delivered — these changes need to keep entering the public corpus. A one-time website revamp cannot solve this problem; continuous information management can.

4. Where We Stand

What Tianxia Gongchang builds is information infrastructure for manufacturing: on one side, we use a database covering 4.8 million real factories in active production to help buyers and upstream salespeople find real factories through AI conversation; on the other side, that database and the industry content built on top of it are also helping real factories get read — and described accurately — by more AI entry points.

If you run a manufacturing business and want to know what you look like in AI's eyes, start with a test: open any AI assistant, ask it for recommended manufacturers in your niche category, then ask it about your company. If the answers fall short, start managing your company's information as an asset today — and if you need professional help, come talk to us through Tianxia Gongchang.