Some Merchants Are Getting 5x Order Growth from AI—Are You Still Stuck in Traditional E-commerce?

Some Merchants Are Getting 5x Order Growth from AI—Are You Still Stuck in Traditional E-commerce?

Have you noticed this frustrating phenomenon?

Your product detail pages are beautifully written—"luxury texture," "premium design," "maximum ambiance." But when you ask Doubao or DeepSeek "what are some good travel bag recommendations," AI never mentions your brand.

Even more frustrating: your competitor's product pages are less polished than yours, yet AI recommends them repeatedly, and orders keep flowing in.

📌 Key Takeaway

36Kr reported today (October 17) that some merchants achieved 5x order growth through GEO optimization on AI platforms like Doubao. World-renowned water bottle brand Stanley saw significant improvement in AI recommendation traffic after optimizing product information; Japanese sports brand Descente achieved a 22% increase in AI-driven recommendation traffic after restructuring their Tmall product pages. The core logic: AI reads products completely differently than humans—humans see prose, AI reads structured data. AirOps analyzed 21,000 brand-related content pieces and found that 85% of AI citation sources come from third-party content, with only 15% from brands themselves. Douyin e-commerce has launched four major upgrades for the Doubao channel, treating AI as an independent business line. GEO is not mysticism—it's quantifiable technical work.

AI Reads Products Completely Differently Than Humans

Many e-commerce operators have a misconception: product detail pages need literary flair, ambiance, and emotional appeal.

AI doesn't care about any of that.

When humans browse product pages, they might be moved by names like "Weekend Vibes," finding them stylish. But AI extracts almost zero information from such names—no category, material, or capacity data.

AI reads structured data with clear field names.

For example:

Material: Organic cotton

Capacity: 40L

Waterproof: Yes

Use case: Airline carry-on, short trips

Large blocks of marketing copy on product pages? AI typically skips right over them.

How Stanley Made AI Understand Their Products

World-renowned water bottle brand Stanley recently optimized product information across almost all channels.

Their Chief Brand Officer stated bluntly: Past marketing centered on Mother's Day and Teacher's Day themes—"more images, less text," focusing on ambiance rather than utility. Easy to move consumers, but incomprehensible to AI.

To make AI understand their products, Stanley did three things:

1. Created FAQ lists for each product

Not generic product descriptions, but organized answers around real questions users ask AI:

Is it suitable as a gift?

Can it be used during workouts?

Can it be brought on airplanes?

Whatever users ask, they answer.

2. Added AI-readable product care instructions and usage guides

How to use, how to maintain—all clearly written. The goal: give AI factual basis when answering consumer questions.

3. Mapped product features to user benefits one-to-one

Let AI directly connect features to benefits when answering users, without guessing.

The result? Significant improvement in AI recommendation traffic.

How Did Descente Achieve 22% Growth?

Japanese sports and outdoor brand Descente restructured their Tmall product pages, resulting in a 22% increase in AI-driven recommendation traffic.

What did they do?

Changed product titles from "Weekend Vibes" to "Canvas Travel Bag, Unisex, 40L, Large Capacity, Water-Resistant, Tan, Airline Carry-On."

Category, material, capacity, function, color, use case—all in the title. AI can instantly grasp all product information.

When consumers ask AI "recommendations for large-capacity waterproof travel bags," this product is much more likely to be recommended.

Looks verbose, but each stacked attribute gives AI one more piece of information.

85% vs 15%: The Brutal Truth About AI Information Sources

This is what most people overlook.

You think AI will cite you just because your brand website is well-made?

Wrong.

Data analytics firm AirOps analyzed 21,000 brand-related content pieces across ChatGPT, Claude, and other AI assistants, finding:

85% of information sources come from third-party content, with only 15% from brands themselves.

Peec AI expanded the sample to 30 million AI-cited content pieces, with an even more concentrated conclusion: the most-cited websites are just a handful—Wikipedia, Reddit, YouTube, LinkedIn, etc.

In China, this translates to platforms like Xiaohongshu, Douyin, Kuaishou, and vertical authority websites.

When AI recommends products, it primarily adopts third-party content, not brand content.

AI Information Source Priority Tiers

Among these third-party platforms, AI has clear source priority tiers:

Tier 1: Central media, government websites, and authoritative academic journals

People's Daily, ministry websites, academic papers

Tier 2: Provincial party media, national industry media, and listed company official sites Tier 3: Major portals, vertical leading media, and industry association sites

Portal tech channels, industry associations—main force for long-tail knowledge

Tier 4: UGC platforms like Xiaohongshu, Douyin, Kuaishou

If your brand information appears on Tier 1 and Tier 2 platforms, your probability of being cited by AI increases dramatically.

Douyin E-commerce's Four Major Upgrades

Platforms have smelled the opportunity.

Douyin e-commerce has comprehensively upgraded for the Doubao channel:

1. Launched Doubao Operations Zone

Separates orders from Doubao channel

Merchants can view AI order GMV, order volume, product card impressions, user clicks, etc.

2. Helps Merchants Find Opportunities on Doubao

Integrates AI channel user behavior and habits before purchase

Such as industry decision factors, frequently asked questions before ordering

3. Helps Merchants Optimize Product Information

Improves product attributes and selling points around high-frequency pre-purchase concerns

4. Sets Up Doubao Channel Exclusive Coupons

Doesn't affect Douyin mall pricing system, accelerates AI channel user decisions

At least within the Douyin and Doubao ecosystem, the platform treats AI channels as an independent business line, not just a supplementary traffic entry.

Three Actionable Points for E-commerce GEO

1. Convert Product Data to Machine-Readable Format

This is the foundation of GEO and the first step for most merchants.

Titles: Category + material + capacity + function + color + use case

Attributes: Fill all fields, leave nothing blank

Descriptions: Structured presentation, minimize marketing copy

2. Pre-emptively Occupy AI-Trusted Information Sources

This is the hardest part of GEO, but also where you can gain the most competitive advantage.

Deploy authentic user content on Xiaohongshu, Douyin, etc.

Seek coverage from industry media and vertical websites

If possible, obtain endorsements from authoritative media

3. Validate Results with Data

Regularly search relevant keywords on Doubao, DeepSeek

Record whether your brand is mentioned, at what position, in what context

Track traffic and order changes from AI recommendations

Final Thoughts

E-commerce traffic logic is undergoing fundamental change.

The old model was "people find products"—users enter keywords in search boxes, browse lists, click to buy.

The new model is "products find people"—users describe needs to AI, AI directly recommends products, users purchase with one click.

If your product isn't in AI's recommendation list, you're excluded from the user's decision starting point.

The good news: GEO is not mysticism—it's quantifiable technical work.

Structure product information, deploy third-party sources, continuously monitor and optimize—three steps, and you too can achieve AI-driven order growth.

Those merchants who've achieved 5x growth aren't lucky—they understood the game rules early.

Are you ready?