---
title: "After the 315 Exposé, Why Did Legitimate GEO Providers See Orders Surge?"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2026-04-07T10:14:13+00:00"
canonical: "https://ffcap.cn/en/research/src-20260407-02html"
source: "https://uniqueresearch.substack.com/p/src-20260407-02html"
language: "en"
---

# After the 315 Exposé, Why Did Legitimate GEO Providers See Orders Surge?

_Original · Unique Research · 2026-04-07_

_Editor's note: This historical edition retains the source writer's narrative and the full panel, including repeated claims and disagreements. “315” refers to the March 15 consumer-rights broadcast discussed in the article. The account of the broadcast, Apollo-9, client demand, research partnerships, industry standards and platform meetings is source-reported, not independently verified here; Apollo-9 is explicitly described as fictional. Search-index movements are proxies, not proof that GEO caused a purchase. The source does not define the denominator of its 90%/95% “revenue-driving traffic” figures or supply an audited attribution model. UTM, checkout, paid-search and platform-data statements describe the speakers' understanding at the time, not current implementation guidance. The source names a luxury client as “L'Oréal (or Estée Lauder)”; that uncertainty remains unresolved. Four-digit fees have no stated currency. The initial January reference is not silently reconciled with the later January ’25 statement. First-person observations belong to the original writer or speaker. Chinese personal and company-name romanizations remain provisional. Provider claims, exclusivity promises and interpretations of compliance are not Unique Research endorsements or legal conclusions. The seven source images and final editorial/duplicate review remain outstanding, so this draft is not approved as a zero-omission public edition._

Extraordinary Awards

The 315 Broadcast Gave GEO a Free Round of Industry Education

After Illicit Practices Were Named, Legitimate GEO Providers Saw Orders Surge

"

Dr. Dong Haoyu, CEO of Yangqi Technology, said something onstage that stuck with me: “Last week, I attended 17 client meetings. Clients are starting to focus on the compliance of GEO services and brands' long-term value in AI search.”

This was the first week after the 315 consumer-rights broadcast exposed illicit GEO practices. In theory, after being called out by CCTV, the industry's normal response should have been anxious introspection behind closed doors. Yet every one of the four panelists said orders had surged the previous week.

That contrast brought the GEO industry's current position into focus.

The 315 Broadcast Gave GEO a Free Round of Industry Education

Yangqi Technology CEO Dong Haoyu began studying GPT's AI search algorithms in January 2023. By this roundtable at Hangzhou AI WEEK, he had accumulated three full years of data and industry insights. In his view, irregular GEO practices such as bulk AI generation, fabricated content, and mass submission to dubious sources began appearing last year. Industry self-regulation and standards urgently need to move onto the agenda.

The central revelation of the 315 broadcast was that unscrupulous merchants used “GEO optimization system” software to mass-produce false advertorials—fabricating product selling points, industry rankings, and user reviews—and automatically publish them online through preset accounts. After large models crawled that content, an entirely fictional “Apollo-9 smart band” entered the top recommendations of mainstream AI models within two hours.

After the program aired, legitimate GEO providers' phones rang off the hook.

Taimei AI CEO Cai Xiaoxu had publicly explained the difference between black-hat and legitimate methods as early as January, but the response was modest. After 315, several clients expected to cooperate before the Lunar New Year said outright on March 16: “Get the contract ready immediately.”

Riven Gao and his team at Yidian Tianxia had acted even earlier: on March 7, in Shenzhen, they issued a GEO industry self-regulation pledge. After 315, it resurfaced and spread widely. His point was direct: “If you do not do GEO, your competitors may capture the AI mindshare positions that replace the traffic positions on the old SERP.”

QuickCreator CEO Zhou Haipeng took a calmer view: “Whenever a channel takes off, there will inevitably be good and bad players. This is a long-term undertaking. It is unlikely to turn entirely black-hat.”

The four emphasized different points but reached the same conclusion: exposing illicit practices had given legitimate GEO a free round of market education.

The Hardest Question: Where Is the ROI?

Moderator Duan Hongyu asked what every brand wanted to know: exposure through AI has increased, but click-through rates fall when a page displays “AI-generated.” Is the money we are spending actually worth it?

The question touched GEO's biggest current weakness.

Cai Xiaoxu explained a basic distinction: GEO results in AI recommending your product. Consumers see an authoritative recommendation, not an “AI-generated” label. They are not the same thing. The deeper problem is that data does not flow between AI models and e-commerce platforms, so you cannot directly see whether Doubao or Qianwen brought in a purchase.

He uses an observable proxy: without adding other marketing activities and with the budget unchanged, check whether the brand search index in Taobao Business Advisor or the search index in Douyin's Business Compass changes after implementing GEO. “If the index changes, I think that shows the logic of genuine business conversion.”

But he also acknowledged that the final purchase cannot currently be verified. His exact phrase was: “Wait patiently for the flowers to bloom.”

It is an interesting phrase. It describes the industry's real condition, as well as a reasonable strategic choice, because he immediately added something practical: “Model traffic is now at a stage where there is a Gap between value and price. It offers the best value for money; now is the most economical time to act. Once it is verified and all your competitors are doing it, costs will rise.”

The Loop Is Forming. You Just Cannot See It Yet.

Dong Haoyu shared a detail that I thought was among the most information-rich moments of the roundtable.

During last year's Black Friday, Amazon traffic was proportional to search traffic on ChatGPT.

Last October, GPT added UTM parameters to its own links. That means a standalone site's traffic dashboard can show directly which visitors came from ChatGPT.

Around this year's Lunar New Year, Yangqi Technology's UK team attended a closed-door OpenAI meeting in London. ChatGPT had a request: it wanted brands to add ChatGPT tracking code to the final transaction page of their standalone sites, closing the loop from GPT-referred traffic to final conversion.

There are similar signals in China. He mentioned that some clients can already place orders in Douyin stores with one click through Doubao. There is simply no data yet to measure the scale of those conversions.

Riven Gao added more specific figures from the perspective of cross-border DTC brands: among the leading brand clients they serve, 90%, or even 95%, of revenue-driving traffic comes from ChatGPT, while Perplexity contributes less than 5%. Grok has a high citation rate but poor commercial value. Leading overseas brands performing well can earn more than US$30,000 a month in revenue from ChatGPT.

Taken together, these figures are not about whether GEO can bring conversions. They suggest that the loop is already forming, with some data still waiting for platforms to unlock it.

Dong Haoyu sees the trend as a move from GEO to GEM—Generative Engine Marketing, meaning paid AI search. “When paid AI search arrives, it will inevitably become an important part of internet advertising,” he said.

If that logic holds, brands doing GEO now are securing the cheapest positions before the era of paid AI advertising arrives.

What Should New Brands Do? What About Established Brands?

There was a practical exchange about the differences between new and established brands.

Dong Haoyu said a new brand's first need is straightforward: Doubao, Qianwen, and Kimi cannot find me. How can you make them find me first? “First solve the problem of going from 0 to 1.”

He described not “making up articles,” but identifying a product's distinctive selling points and amplifying them through sources visible to AI search. For example, during a product launch, they helped a luxury brand move from zero to visible in the search scenario “luxury-brand lipstick for my girlfriend at Qixi.”

The fees can be low, “affordable in the four-digit range,” but “the value of intellect, strategy, and algorithms is beyond price.”

Cai Xiaoxu's perspective was closer to strategy consulting. A new brand's central problem is insufficient consumer awareness. The solution is to find the smallest use case and go deep. Air purifiers are already a fiercely competitive category, but if you focus on the specific selling point “small size, large coverage” and embed it deeply in AI recommendations, “the market is big enough if you serve it deeply enough.”

Riven Gao offered a more systematic framework. They divide GEO services into three layers:

First, the AI visibility layer. Look not only at whether a brand is mentioned, but at mention frequency, whether it is recommended first, referral-link quality, and whether the AI-generated answer expresses positive or negative sentiment toward the brand. Even user reviews from three years ago can influence AI's judgment.

Second, the traffic layer. Track the traffic AI platforms bring and its quality. ChatGPT now provides UTM parameters, and Google Gemini also provides parameters; AI Overviews does not yet.

Third, the Revenue layer. But he emphasized that it cannot be the only measure: “If you look at just one, the project becomes distorted.” GEO occupies mindshare in AI answers during consumer decision-making, so brand presence and ROI must be considered together.

Zhou Haipeng's advice was more pragmatic, even sobering: get SEO right first, because a great deal of traffic still comes through search engines. Treat GEO as brand building, with process indicators such as brand presence, comparative recognition, and sentiment.

His example was specific: for CATL, there is a highly specialized European electric-vehicle industry forum called EV Magazine. “That is where your real B2B customers gather.” X and Facebook provide a layer of broad exposure, but genuine endorsement comes from specialist media. For to B products, expert responses are necessary to create real endorsement.

Marketing Still Begins with the Consumer Decision Journey

Toward the end of the roundtable, Dong Haoyu said something I thought was exactly right:

“Marketing must always return to the consumer decision journey.”

From AIPL to 5A, consumers may ask AI at every decision stage: first, “What product do you recommend?” Then, “How well does this product work?” Finally, “Where can I buy it more cheaply?” If you appear at none of AI's touchpoints, you are not on the consumer's path to a choice. It is like a brand failing to build a mini-program or open a Douyin store in the earlier era.

More than 300,000 new SKUs appear in China each year. Every new product needs a website, a mini-program, and an official account. Adding AI search as another touchpoint is not merely a bonus; it fills a missing link from the mobile-internet era.

The GEO loop is not fully connected yet. Doubao and Douyin-store data are not completely aligned, and GEM—Generative Engine Marketing—in China is still in internal testing. But as Cai Xiaoxu said, once platforms connect the data, verification will be obvious. The only question is when you want to start positioning yourself.

More Details from the Conversation

Extraordinary Awards · Hangzhou AI WEEK Trends Roundtable Panel: “Visibility Marketing: From Traffic Acquisition to Business Conversion”

Guests: Dong Haoyu, CEO of Yangqi Technology | Cai Xiaoxu, Founder and CEO of Taimei AI | Riven Gao, Head of DTC Brand Technology at Yidian Tianxia | Zhou Haipeng, Founder and CEO of QuickCreator

Moderator: Duan Hongyu, Partner at Unique Research

Duan Hongyu: I am Duan Hongyu, a partner at Unique Research. Today's four guests represent some of China's leading GEO companies. As everyone knows, GEO was a hot sector within AI before, but since 3.15 it has been at the center of a storm. So let us first ask our four guests to briefly introduce their core businesses and explain how they view perceptions of the GEO market following the 315 broadcast.

Dong Haoyu: I am Dong Haoyu, founder and CEO of Yangqi Technology. I have always defined our business as an AI search-algorithm and GEO-agent company. So our foundation remains algorithm-driven: helping brands improve their visibility in AI, as today's theme puts it, and then generating stronger consumer intent to convert. We began researching GPT's search algorithms and recommendation logic in January 2023. We have now been doing it for about three years and have accumulated a little data. Last year, together with Peking University, we published what we called China's first technical methodology, Stream. People in this industry generally know the Princeton methodology. China also has a technical methodology developed jointly by Peking University and Yangqi Technology. We have always believed in trustworthy communication through AI. I think that is very important.

We also saw that the 315 report covered black-hat GEO practices. After the report, many companies learned that GEO now has illicit and gray-market activities. So last week our orders surged. I personally attended about 17 client meetings that week, and I believe everyone here was in a similar position. We therefore believe exposing illicit and gray-market practices is actually a very positive development for legitimate, trustworthy GEO communication. Yangqi Technology is also working with the China Advertising Association on industry standards for GEO in China. Legitimate, compliant, trustworthy AI communication is the right path for GEO from here.

Cai Xiaoxu: I am Cai Xiaoxu, CEO of Taimei AI. We are actually similar to Yangqi: we entered AIGC comprehensively in ’23, focusing on marketing. Over the course of ’23, we deployed dozens of marketing-focused agents covering areas such as public-sentiment insights and comment operations. Ultimately, in response to customer demand, GEO became a necessary service. We also fully embraced GEO in ’24 and ’25, gradually serving dozens of clients across five major industries.

By January ’25, we had already seen some providers using black-hat methods that damaged the industry's rules or ecosystem. At that point, we also published articles publicly to help people distinguish black-hat methods from legitimate ones. Then the 315 exposé happened, and people finally understood what legitimate methods looked like and what we did. Our experience was the same as Mr. Dong's. We were incredibly busy last week. Beyond meetings, several clients we expected to work with around the Lunar New Year said on March 16: “Get the contract ready immediately.” So as long as providers use legitimate methods and deliver genuine visibility, business will not be bad.

Riven Gao: I am Riven from Yidian Tianxia. We are a company focused on overseas marketing. My department mainly provides integrated SEO and GEO marketing-technology solutions for cross-border DTC brands, helping them implement the entire path from visibility to conversion. I personally moved into GEO from Shopify and SEO. Since ’19, before the pandemic, I have been studying Google-related SEO strategies and underlying algorithms for overseas markets, mainly serving standalone-site clients.

Since the 315 exposé, we have felt that the industry had indeed been chaotic. On March 7, Yidian Tianxia held a visibility summit in Shenzhen, bringing together ecosystem partners to issue a GEO industry self-regulation pledge—a way for good practices to drive out bad ones. As the other business leaders just mentioned, since 315, more people have paid attention to GEO. For DTC brands, it is something they have to do. If you do not do GEO, competitors may capture the AI mindshare positions that replace traffic positions on the old SERP. This needs to be a priority.

Zhou Haipeng: I am Zhou Haipeng from QuickCreator. Our company may differ a little from the others: we work only on overseas markets. Our customers include the substantial foreign-trade communities of eastern China's Jiangnan region and southern China, as well as a to C segment. As they compete overseas, GEO has become an indispensable channel alongside the SEO we did before. So in 2026, we hope to help everyone build credible brand recognition across global channels. That is something both service providers and their clients should work on over the long term.

As for the 315 issue, I think it is normal. Whenever a channel takes off, especially with capabilities now so powerful, there will inevitably be good and bad players. But over the long term, something like GEO is an undertaking for an entire era. It is unlikely to turn entirely black-hat. With many clients and providers working together, I believe it remains a relatively fair market. You can rest assured: 315 was certainly not exposing people like us. So I think we should keep working to improve this market and help Chinese businesses expand overseas and build stronger global brands.

Duan Hongyu: It sounds as though 315 has actually educated the GEO market. But while learning about GEO and speaking with brands, I have encountered a lot of feedback online: after using AI for SEO and generating advertising content, exposure increases, yet click-through rates fall when the page shows that the content was AI-generated. How do you see that? Let us start with Mr. Cai from Taimei.

Cai Xiaoxu: First, what GEO ultimately produces is what we call mention rate: the recommended answers include our relevant brands and products. When people then click through to articles across media channels, those articles generally will not display an AI-generated label. Returning to conversion, we need to understand that GEO ultimately gives AI a basic understanding of our product, service, or industry. Once it has that understanding, it recommends those services and products to consumers in an authoritative way. So it is more about establishing trust in consumers' minds. With that trust, consumers are more likely to buy your product when they make a purchase decision.

On conversion rates, the problem is that connections between models and major e-commerce platforms have not been established. That makes it difficult to verify conversions through a clear, closed data loop. But there are some third-party verification paths. After doing GEO, for example, you can look at the brand search index in Taobao Business Advisor or the search index in Douyin's Business Compass. Without adding any other marketing activity, with the budget unchanged apart from adding the GEO budget, has your brand search index increased? If it changes, I think that shows the logic of genuine business conversion. It is just that at the final purchase stage, we cannot currently verify whether Doubao, Qianwen, or Kimi brought it in. The platforms' data is not connected; if it were, we could verify it. For now, we call it “waiting patiently for the flowers to bloom.” Once a platform opens that data, verification will be very clear.

But we cannot say that just because we cannot see conversions now, we should do nothing. Model traffic is now at a stage where there is a Gap between value and price. It offers the best value for money; now is the most economical time to act. Once it is verified and your competitors and outside brands all join in, costs will rise. So we need to face this clearly: first, exposure can certainly be achieved; second, getting target consumers to trust our products is most important. Then look at conversion.

Duan Hongyu: Mr. Cai, let me ask one small follow-up. You said it is currently difficult to describe platform data clearly in quantitative terms. Under these circumstances, how do we balance AI efficiency with genuine commercial value? Are there other ways we might quantify it?

Cai Xiaoxu: In measurement, we still need to look at brand value. Today, when all these apps have more than 100 million monthly active users, the greatest value is whether your target consumers can encounter, recognize, and trust you on such platforms. As an aside, through visits to many brand clients, we found that from ’24 to ’25, all the brand clients deeply engaged with Douyin were losing money if you looked at ROI alone. Why did they keep investing in Douyin? Because they saw its short videos and livestreams as brand exposure, with conversion providing additional incremental gains. So commercially, I think that scale of brand exposure is the greatest value.

Duan Hongyu: Mr. Dong from Yangqi Technology, how do you see it?

Dong Haoyu: Here is how I see it. From a marketing perspective, there are two paths: AIPL—Awareness, Interest, Purchase, Loyalty—and the more familiar 5A. Along a consumer's brand-selection journey, AI search plays a role at every decision stage. For example, a consumer might ask for a smart-glasses recommendation based on their personal circumstances. That is the first step, Awareness. Once interested, they may ask about the product's functionality and how it compares. Third, they may ask where it is cheaper to buy.

Last year, we saw ChatGPT data showing that during Black Friday, Amazon traffic and search traffic on ChatGPT were essentially proportional. And last October, GPT added UTM parameters identifying ChatGPT to its links. The backend of a standalone site can already show very clearly that traffic comes from ChatGPT, so we see that, in a sense, a loop is already forming overseas. Around this year's Lunar New Year, Yangqi Technology's UK team attended a closed-door OpenAI meeting in London. One thing requested there was for brands to add ChatGPT tracking code to the final transaction page of their standalone sites, completing the loop from GPT-referred traffic to final conversion.

I believe we will see that conversion loop emerge within the next year or two. For example, while speaking with Doubao's commercialization team about what the future might look like, we found that although the backend data does not yet appear connected, some of our clients can already place orders directly in Douyin stores with one click through Doubao. We simply cannot yet measure the value of those conversions. As the loop forms quickly, GEO becomes GEM—Generative Engine Marketing, meaning paid AI search. When paid AI search arrives, it will inevitably become an important part of internet advertising. So the future AI search market will be made up of GEM as a whole, and people may then see ROI rise rapidly and become highly correlated with it.

An important point is that if a brand cannot appear on AI, it is not on the consumer's initial path to a choice. Some overseas TikTok Shop brands do not appear on ChatGPT, so some have begun positioning themselves there. Likewise in China, consumers ask how Taobao-native and Douyin-native brands can appear on Qianwen, Doubao, and Kimi. I think marketing must always return to the consumer decision journey.

Duan Hongyu: Mr. Gao and Mr. Zhou, since your companies mainly work on overseas markets, let us focus on that context. In overseas GEO, how do you balance the exposure we have discussed with eventual commercial conversion and ROI?

Riven Gao: The team we lead at Yidian Tianxia mainly focuses on integrated SEO and GEO marketing for cross-border DTC brands. In the Cases we have served, our solution has three dimensions of metrics.

The first is the AI visibility layer. We use corresponding Keywords or Prompts to examine visibility across major overseas AI platforms such as ChatGPT, Perplexity, Gemini, and Claude. We look not only at whether a brand is mentioned, but how often, whether it is recommended first, whether the referral links below include your onsite content or offsite PR, and whether the generated answer expresses positive sentiment toward your brand. If the brand has data on UGC platforms such as Reddit, or Trustpilot reviews, even user reviews from three years ago can influence AI's judgment.

The second is the traffic layer. As Mr. Dong mentioned, ChatGPT added UTM parameters, so in the Agent Commerce era we can track this traffic. Google currently provides parameters only in Gemini, not AI Overviews. At the traffic layer, we can see a “partly closed loop” from GPT. They are already working on Instant Checkout and connecting payments such as Stripe, and in future Agent Commerce could close the loop fully. We also need to examine traffic quality. In our monitoring, leading DTC brand clients receive roughly 90%, or even 95%, of revenue-driving traffic from ChatGPT; Perplexity may contribute less than 5%. Grok has a high citation rate, but its actual commercial value is not very good.

The third is the Revenue layer. GEO cannot currently be measured only by revenue. If that is the sole metric, the project becomes distorted. We are occupying mindshare in AI answers throughout consumer decision-making, so we need to consider Branding and ROI together. Looking at broader data, leading overseas brands performing well can earn more than about US$30,000 a month from ChatGPT. Let me add some industry observations: I am very optimistic about GEO Commerce overall. GEM is still in internal and public testing. Once ChatGPT succeeds in exploring it—one reason Google has not shifted large volumes to AI Overviews is that doing so would hurt its own SEM revenue—it must first make GEM work to carry out its own revolution, connect payment ecosystems with Shopify and others as quickly as possible, and synchronize Google Merchant Center (GMC) data.

Zhou Haipeng: Let me continue on conversion. The first layer is conversion from market-level brand exposure to actual site visits. That is what GEO does in most cases. Whether from a brand perspective or by targeting large volumes of long-tail keywords, the aim is broad market exposure. But conversion from exposure to site visits is quite difficult to control, because it is not like seeing an ad and clicking immediately. If you build content marketing around long-term brand development, conversion rates can be hard to perceive at times. Yet many successful overseas companies grew this way, such as Canva and Notion. There is also a very low-profile but excellent software company in China that has consistently produced content for years and now receives a great deal of traffic every day. This is an accumulated result that requires patience.

The second layer is conversion after users arrive on the site, which largely depends on your own execution. From registration to clicks and purchases, and even to B after-sales service, I think many mature tools, systems, and operating models already support this in China and overseas. The difficulty now is the dramatically accelerating pace of iteration. There may once have been ten providers of data-analysis tools; now there may be 100. Meeting customer needs and using the product well depend on your own capabilities. So the first issue is a marketing issue; the second is a product-operations issue. Separate them. A business leader may have different KPIs for each, and that is how the path from GEO to cash can be connected.

Duan Hongyu: You mentioned that there are many complex channels overseas. To put it bluntly, for most brands trying to validate their approach, if they focus on standalone sites plus growth across platforms, which might perform better from an ROI perspective?

Zhou Haipeng: I actually think specialized channels are better, whether for ROI or brand endorsement. People often glance at broad information feeds and leave. The genuinely valuable places are specialized third-party channels. For example, after providing a service for CATL, we discovered many channels we had not noticed before. In Europe, there is a relatively niche electric-vehicle and battery-industry forum called EV Magazine. Those are the places where your real business customers gather. Broad media such as X and Facebook provide one layer of exposure, but those audiences naturally congregate in specialist media. Such media are both a place for you to speak and a source of endorsement, greatly improving recognition.

For to C products, it is often the public's response that matters. For to B products, expert responses are necessary to create genuine endorsement. Some third-party professional media in China may currently be weakening—for example, certain medical journals—but I think independent third-party media in specialized fields will play an increasingly important role.

Duan Hongyu: Let me ask Mr. Dong another question. When a new brand invests in visibility marketing, do you set any expectations—for example, how much it invests before it may need to add some products?

Dong Haoyu: For a new brand, the first step is strengthening awareness and exposure. We serve quite a few new brands whose first request is: “Doubao, Qianwen, and Kimi cannot find my brand. How can you make them find me first?” It is about going from 0 to 1. That also relates to brand character, the founder's style, and the product's unique selling points. GEO is not casually inventing an article. It means identifying distinctive selling points and amplifying them through sources visible to AI search.

I discuss with founders what the core selling point is: emotional appeal or product strength? At Yangqi Technology, we use algorithms and agents to help clients identify differentiation, then develop AI search strategies, distribute content, and monitor results. If the differentiation is sufficiently clear, fees can be very low, affordable in the four-digit range. But the value of intellect, strategy, and algorithms is beyond price.

Also, it is not only new brands that need GEO; established brands' new products do, too. Last year, for example, we helped L'Oréal (or Estée Lauder) with GEO during a new-product launch, taking it from zero AI visibility to one and capturing a scenario such as “luxury-brand lipstick for my girlfriend at Qixi.” More than 300,000 new SKUs appear in China each year, and visibility comes first. Previously, a new product started with a website, then mini-programs and official accounts. In the future, new products will also need AI search. AI search fills in a touchpoint missing from the mobile-internet era. So I always tell clients that they need a new line in their advertising budget called AI search.

Duan Hongyu: Mr. Cai, on Mr. Dong's point about established and new brands, let me first ask: are most of Taimei's paying clients established brands or emerging ones?

Cai Xiaoxu: In absolute proportions, definitely more established brands and fewer new ones.

Duan Hongyu: How do you think new brands can establish an advantage against the accumulated strengths of established brands? What do you do?

Cai Xiaoxu: A new brand's core problem is insufficient consumer awareness. In the past, building that awareness required enough money for Focus Media's building-screen ads, Xiaohongshu discovery marketing, or broad Douyin exposure. Without that budget, a new brand needs to find an undervalued opportunity and go deep in a scenario that has not been sufficiently segmented. For example, air purifiers are fiercely competitive in China. How do you find the smallest use case? Go very deep into it so consumers recognize you in that specific context. When we develop strategies for new brands, we differentiate positioning around the initial intent questions, then go deep and broad once we identify those core questions. For an air purifier, for example, we embed the differentiated concept “small size, large coverage” deeply into AI. Whenever consumers search related questions, it recommends only that brand. If you serve this market deeply enough, it is big enough.

Duan Hongyu: Because of time, could Mr. Gao and Mr. Zhou each answer one final question briefly? In 2026, if a brand wants to go global, what should be the most important value metric for assessing its conversion?

Riven Gao: From our perspective, the core of GEO is working on “entropy.” A brand exists in a chaotic state within AI, requiring contextual information to be processed and output across different models. The most important thing is “entropy reduction”; that is the core metric. We develop GEO monitoring tools. On one hand, a human team works on Reddit, PR, and onsite optimization. On the other, we Scale the GEO capabilities used for major clients, building an Agent Team to automate operations, lower costs, and improve efficiency. In one sentence: understand entropy and reduce it for AI, with Agents doing the work.

Zhou Haipeng: Let me add something. In many cases, GEO's foundation is getting SEO right first, then doing GEO. Large language models and their business models are not yet fully developed. From a Marketing department's overall perspective, most small and new brands should get SEO right first, because a great deal of traffic still comes through search engines. Google itself combines search and AI, with clear methodologies and measurement systems. Then treat GEO as brand building, using process indicators such as brand presence, comparative recognition, and sentiment. In a business leader's mind: get SEO right first, then view GEO as brand building.

---

Original publication: https://uniqueresearch.substack.com/p/src-20260407-02html
On-site reading page: https://ffcap.cn/en/research/src-20260407-02html