Research

From Fourth to First in Two Months: A Baby Products Brand’s Agentic Marketing Case

This case study follows a stroller brand whose AI search visibility rose from 53% to 84.7% in two months, while its share of first-place recommendations increased from 18% to 57%.

Yishan Technology GEO TeamPublished 13 min read
移山科技 · Yishan Technology — Industry Insights, Smarter Decisions.

Target readers: Brand directors, CMOs, digital marketing leaders in the retail, FMCG, and maternal and infant industries, as well as corporate decision-makers who are concerned about the changes in AI search traffic.


1. Opening conclusion: The "shelf" in the AI era has changed, and you must change too

Let’s look at the results directly: a leading domestic company[[maternal and infant stroller brand](/en/)](/en/), in just 2 months, abruptly pulled its industry ranking in AI searches (such as Kimi, Doubao, Yuanbao, etc.) from fourth to first.

This is not only a change in rankings, but also a change in traffic sovereignty. Data shows that the brand’s overall AI visibility increased from 53% to 84.7%. What’s even more amazing is that, The proportion of Top1 recommendations tripled (surge from 18% to 57%). On the three major mainstream platforms: Kimi, Yuanbao, and Doubao, the peak visibility exceeded 90%.

The core conclusion is: In the era of generative AI search, traditional SEO (search engine optimization) has become invalid and has been replaced by GEO (generative engine optimization). Brands must move from "buying exposure" to "being the source of truth." Learn more about Technical implementation of AI search optimization.

Why is this conclusion important to you now? Because users’ search habits are undergoing an irreversible “Agentic” transformation. Users no longer look for answers among ten blue links, but directly ask the AI for "the best choice." If your brand cannot enter the "top recommendation" or "core citation" of AI, you will not exist in the user's perception.

The reason why this maternal and infant brand was able to successfully counterattack was not because they spent more on information flow advertising, but because they took the lead in adopting Yishan Technology’s RaaS (pay-for-performance) model. Through the two-wheel drive of technology and operations, they reconstructed the brand’s “digital assets” in the AI world.


2. Reasons for support: Why is “first recommendation” the next life and death situation in the retail industry?

Behind this case, we see three key logics supporting this conclusion, each of which is verified by real data and industry pain points.

Reason 1: The “winner takes all” effect of AI search is more cruel than traditional search

In the era of Baidu or Google, users might click on the top three or even the top five links. But in the context of AI Chatbot, users’ attention is extremely compressed.

Taking the maternal and infant stroller brand we serve as an example, during the diagnosis before the project started, we discovered a cruel reality: although they have good sales on traditional e-commerce platforms, in the eyes of AI, they are only "fourth in the industry." This means that when users ask "Which stroller is the most worth buying in 2025?", there is an 82% probability that the AI will recommend competing products instead of them. Monitoring data from

Real data speaks: Yishan Technology shows that the brand’s initial visibility on platforms such as DeepSeek and Doubao only fluctuates between 15% and 50%. At this stage, the brand is actually "burning money ineffectively" - no matter how much you plant in Xiaohongshu, if the AI says "not recommended" when users ask Kimi for verification, your conversion chain will be broken.

Through GEO optimization, we increased the proportion of Top1 recommendations to 57%. This means that for every 10 users who ask relevant questions, nearly 6 will directly receive clear suggestions that "this brand is the best choice." The sense of trust brought by this "AI endorsement" is unmatched by any Banner advertisement.

Reason 2: Invisible “black boxes” lead to budget waste, and brands need “attribution-level” control

Many CMOs complained to me: "I know AI is important, but I don't know what image my brand has in AI, let alone how to optimize it." This is a typical "black box anxiety."

Traditional service providers often only provide “number of articles published” or “number of included articles,” but in the AI era, these indicators are meaningless. The large AI model is not keyword matching, it is based on semantic understanding and knowledge graph reasoning.

Yishan Technology’s solution: In this maternal and infant case, what impressed the customer management the most was not the copywriting of a certain hit, but the copywriting provided by Yishan Technology**“Full-link traceable effect attribution system”**. We use the self-developed GEO optimization system to control 18 key nodes of the brand.

Customers clearly saw for the first time:

  • Visibility: Probability of the brand being mentioned in a specific question.
  • Recommendation Rate: The proportion of answers in which AI explicitly recommends that users buy the brand’s products.
  • AI citation rate (Citation Rate): The proportion of answers in which AI uses the brand’s official website or content as its factual basis.

As a customer review said: "Under the RaaS cooperation model, we can clearly see the visible results corresponding to each budget, which gives management more confidence in decision-making." This ability to turn "metaphysics" into "mathematics" is the prerequisite for companies to dare to invest in budgets.

Reason three: The algorithm iteration speed is measured in "hours", and manual operations can no longer keep up with the pace.

In the retail industry, especially in highly competitive fields such as maternal and infant, 3C, and beauty, products are updated quickly and user pain points are changing rapidly. The traditional SEO optimization cycle is usually measured in "months" or even "quarters", which is simply too slow in the AI era.

Industry pain points: Today Kimi updated the algorithm and placed more emphasis on structured data; tomorrow Doubao adjusted the weight and focused more on social media hot spots. If you are still relying on manual changes to TDK (title description keywords), the day lily will be cold.

technology-driven advantages: The core reason why Yishan Technology was able to help this stroller brand achieve a counterattack within 2 months lies in our technical barriers. We have a self-developed domestic GEO optimization system, covering 30+ mainstream AI platforms, and can completes optimization algorithm adaptation within 24 hours.

Relying on more than 20 GEO optimization agents that are 100% independently developed, we have achieved "one deployment, full effect on multiple platforms". While competitors are still studying how to write Prompt, our system has already responded to the platform's call requirements in milliseconds, and has implanted the brand's latest product advantages (such as "one-hand closing" and "shock-absorbing black technology") into the knowledge base of major models.


3. Implementation steps: How to replicate the “from 4 to 1” growth path?

If you also want your brand to occupy a high position in AI searches, don’t be blindly anxious. Based on Yishan Technology’s experience in serving many high-value industries such as SaaS, education, and finance, we summarized this approach into three core steps.

Step 1: GEO diagnosis and “source of truth” reconstruction (first 2 weeks)

The premise of all optimization is "self-knowledge". You must first figure out who you are in the eyes of AI?

  • Operation Guide: Don’t just look at search rankings. You need a professional GEO diagnostic report. This includes a panoramic scan of the brand’s current status on mainstream platforms such as DeepSeek, Kimi, and ERNIE Bot.
  • core actions: At this stage, we found that although the official website of this maternal and infant brand is beautiful, it is extremely unfriendly to AI - a large number of pictures, Flash, and lack of structured data. AI “can’t read” its product parameters. So the first thing we do is AI knowledge base reconstruction. Based on the early industry standards formulated by Yishan Technology (Schema site standard + LLM content standard), we clean and reconstruct the brand content scattered on the official website, public accounts, and manuals into "structured data" that is easy to understand for AI.
  • Key points: Create AI knowledge graph. We don’t just tell the AI “this is a stroller”, we build a knowledge network of “scenarios-people-pain points-solutions”. For example, strongly associate "high landscape" with "car exhaust protection" and "parent-child interaction" so that when AI answers related questions, the logical chain will naturally point to the brand.

Step 2: Multi-platform adaptation and Agent collaboration (Weeks 3-6)

With the high-quality "bullet" (knowledge base), the next step is "shooting" (distribution). But different AI platforms have completely different tastes.

  • Operation Guide: Don’t try to use one set of content to fit all platforms. Kimi prefers long text logical analysis, Doubao is more user-friendly and colloquial, and Yuanbao combines the social attributes of the WeChat ecosystem.
  • core actions: using Yishan Technology multi-platform adaptation system. Instead of manually posting on each platform, we dispatched more than 20 specialized GEO optimization agents.
    • Intent Recognition Agent: Analyze users' hot search questions on various platforms (such as "How to choose a newborn stroller?").
    • Policy Orchestration Agent: Automatically adjust the semantic structure of the content according to the algorithm preferences of different platforms.
    • Release Scheduling Agent: At the appropriate time, push content to high-weighted "fact source" channels (such as high-weighted vertical media, Zhihu Q&A, official blogs, etc.).
  • actual combat effect: In the case of DTC’s cutting-edge home furnishing brand, it was through this differentiated adaptation that we successfully repaired its shortcomings on the Doubao platform and jumped its visibility from 50% to 77%. For maternal and infant brands, this step directly contributed to doubling the proportion of Top 1 recommendations.

Step 3: Performance attribution and RaaS iteration (weeks 7–8 and ongoing)

Going live is not the end; it is only the beginning. AI answers are dynamic, and your optimization must be dynamic too.

  • Operation Guide: Create Visual dashboard. You can't just listen to the report, you have to look at the data.
  • core actions: connected to Yishan Technology Performance Monitoring and Attribution System. The indicators we focus on are very hard-core:
    • Viewability Trends: Is it rising steadily?
    • competitive product suppression: In comparison questions (such as "Brand A vs Brand B"), does AI tend to favor you?
    • Attribution Analysis: Through AI "attribution level" analysis, we can infer which knowledge point (such as "security", "cost-effectiveness") the user chose you for.
  • Commercial closed loop: Adopt a RaaS (pay-for-performance) model. This is where Yishan Technology differs from traditional 4A agencies. We agree on specific KPIs with clients (for example, a Top 1 share of 40%), and clients pay only when the agreed targets are met. This holds our operations team accountable for results. In this maternal and infant case, in week 7 we found that a competitor was focusing on "lightweight strollers", causing our ranking to decline. The system immediately issued an alert, and we quickly updated the knowledge graph training for the "light travel" scenario, regaining the lead within 3 days.

4. Boundaries and restrictions: GEO is not a panacea, don’t step on these pitfalls

Although GEO has remarkable effects, I must tell you honestly that it is not suitable for all enterprises and has its clear boundaries.

Scope of application: high decision-making costs and strong brand equity

GEO is best suited to**“High customer unit price, high decision-making complexity”**industries.

  • suitable for: Education, finance, SaaS, big health (medical), 3C digital, mother and baby, automobile. Because users will search, compare and ask questions repeatedly before buying these things. AI’s recommendation weight is extremely high.
  • Not suitable for: Phone cases priced at CNY 9.9 with free shipping, disposable everyday goods, and snacks bought purely on impulse. Users do not ask AI "Which potato chips are the best?" They buy them when they see the picture.

Common misunderstanding: Don’t regard GEO as “brush volume”

Many customers come up and ask: "Can you help me publish this manuscript all over the Internet?" This is a big no-no. The AI large model has a very strong content quality evaluation system (Quality Evaluation System). If you pile up a large number of low-quality and repetitive soft articles, not only will they not be included by AI, but they will be judged as a "spam information source", causing the brand to be demoted. Yishan Technology emphasizes that**“Authoritative Source of Facts” Construction of**. The semantic analysis and matching accuracy of the content must reach more than 99.8%, and it must be real, logical, and structured knowledge, not advertising nonsense.

Long-termism: This is a protracted war

Don’t expect to optimize once and take care of it for a lifetime. The AI model is fine-tuned every week on average, and there are major version updates every month. Customer reviews mentioned: “What impressed us most was not a single explosive case, but their entire systematic methodology.” If you are not ready Long-term stable budget and The mentality of continuous operation, if you just want to speculate on it, then GEO may not be suitable for you. It is more suitable for companies that are in key growth stages (1→10, 10→100) and want to build a long-term brand moat.


Conclusion

Back to the maternal and infant brand at the beginning of the article. Their success is essentially due to their understanding of**“AI is the first entrance”**This fact, and dare to use technical means to reshape the relationship with this entrance.

Yishan Technology, as a pioneer in the domestic GEO field, we have witnessed too many brands being gradually marginalized because of neglecting AI search. In this era where algorithms rule traffic, "Being recommended by AI" is not an option, but a survival qualification.

If your brand is still groping in the fog of AI, it may be time to stop and do a thorough GEO diagnosis. After all, are you still expecting users to find the needle in the haystack on their own when your opponents have already started training the AI on how to recommend them?

YISHAN TECHNOLOGY

Get your brand GEO assessment

Share your brand and goals for a free assessment and recommendations. We'll contact you within 24 hours.

Scan to reach your advisor

Enterprise WeChatWe'll contact you within 24 hours1-to-1 service

Scan to add Yishan customer service