Response Spaces Evolving into Ad Spaces

By Li Xuanqi

Edited by Chen Dengxin

Typeset by Annalee

AI is revolutionizing modern workplaces and contemporary lifestyles. Once upon a time, people would turn to search engines for answers; now, they ask AI. Whether it's ChatGPT, DeepSeek, Doubao, or Yuanbao...

According to Toubao Research Institute, the global user base for AI search tools has surged from 310 million in January 2024 to 1.98 billion in February 2025, marking a staggering growth rate of 538.7%.

As AI increasingly becomes indispensable in daily life, subtle changes are occurring. Brand names frequently appear in DeepSeek's responses, Doubao incorporates ad videos, and Yuanbao's answers feature platform hyperlinks. On social media, users are questioning, "Are there ads in DeepSeek's answers?" Meanwhile, the advertising and marketing industry senses new opportunities: how to make brands stand out and rank higher in AI-generated responses, thereby precisely reaching target users and achieving efficient conversions?

Are AI-Generated Responses Becoming "Ad Spaces"?

Zhao Xinting, a millennial, now has a mantra: When in doubt, ask AI first. From trivial matters to critical work decisions, she relies on DeepSeek or ChatGPT. Recently, concerned about her boyfriend's skincare routine, she asked DeepSeek, "How should men start a skincare routine from scratch?" While reading the thoughtful response, she wondered, "Are these recommended brands really ads?"

The response, spanning cleansing, moisturizing, and sunscreen, listed specific brands and products. For cleansing alone, five brands, including Curél, Mentholatum, and Ponds, were mentioned.

This wasn't the first time Zhao had such doubts. When asking about improving concentration, DeepSeek suggested an app called "Forest," leading Zhao to speculate that it might have been captured due to the app's online marketing. However, she also began to worry about potential ad placements in AI responses.

These doubts are widespread. On platforms like Xiaohongshu and Weibo, users have questioned whether AI responses are turning into "ad spaces." Some mentioned ads in Doubao's responses, while others noted Tencent's Yuanbao suggesting platforms like "58 Home Services" for moving and "Quick Recycling Platform" for recycling, with direct links to their websites. These instances suggest AI responses are "stuffing ads."

Furthermore, users have posted that when searching for a Hong Kong travel guide on DeepSeek, it provided a local tour guide's phone number, which was indeed found on WeChat. "This ad placement is too obvious."

Some speculated that this occurs because AI models integrate data from various sources, leading to "hallucinations" from training data. However, Zhang Jun, Tencent's Director of Public Relations, responded, "Theoretically, it should not be the case. We've asked the technical team to investigate." He suggested asking more questions to determine if it's an ad. Later, Zhang clarified that the answer referenced content from a public account, likely enabling internet search.

During a financial report conference call last November, Tencent's Chief Strategy Officer stated that Yuanbao would not embed commercial search results, focusing on maximizing user appeal rather than monetization.

Advertising on AI Platforms is More Evident

While brands appearing in AI responses are merely suspected ads, advertising on AI social platforms is more apparent. College student Taotao discovered ads on the AI social app "Xingye." When chatting with her usual AI agent, it suddenly mentioned a game and provided a download link, making Taotao feel "very out of place." Similar situations have been mentioned on social media.

Who Smelled the Business Opportunity First?

While users resist ads in AI responses, brands and the marketing industry see new opportunities. One significant manifestation is GEO becoming a buzzword in advertising and marketing.

GEO (Generative Engine Optimization) is a new marketing approach tailored for AI chat models. It involves creating content AI prefers and distributing it on platforms AI likes, subtly increasing the likelihood of business names appearing in AI answers. The Indian Institute of Technology first proposed the concept in the paper "GEO: Generative Engine Optimization," stating that while AI generation is a probabilistic black box, answers are not random. Understanding AI's "content preferences" can influence its output.

SEO (Search Engine Optimization) from the search engine era corresponds to GEO. Previously, SEO supported a nearly $90 billion market.

Today, many marketing companies focus on GEO services. Most transitioned from SEO services to AI search optimization, emphasizing the AI search trend and claiming to "completely change how your business is discovered." Some even position themselves as "DeepSeek promotion companies."

A GEO service company founded in 2015 introduced that it can optimize brand exposure through DeepSeek's recommendation, ranking, and Q&A formats, presenting relevant cases. Their GEO optimization process includes demand analysis, precise intent recognition, landing detection, effect summary, and AI improvement. The main approach is optimizing keywords highly relevant to the business, integrating data from multiple platforms, identifying high-conversion long-tail keywords, and optimizing keywords based on search volume/competition ratio (KCI index).

Most GEO service companies' quotes are linked to the number of keyword entries. The aforementioned company charges 6,000 yuan per month for one root word with 50 entries, including basic service and performance evaluation fees, guaranteeing a 60% AI platform search entry completion rate. Costs increase with more root words and keyword entries, reaching 20,000 yuan for 500 entries.

Another GEO service company mentioned strategies like structured content, highlighting key information, enhancing semantic relevance, continuous monitoring, and adjustment. Structured content optimizes websites and content for easier parsing and referencing by generative engines, including clear titles, subtitles, and meta tags. Enhancing semantic relevance involves using keywords and phrases to align content with target audience search intent.

Their fee structure is also monthly, ranging from 3,800 yuan to 5,000 yuan, depending on the number of keywords and citation effect.

Regarding GEO effectiveness, Xin Kedu consulted multiple providers. Some promised 100% AI recommendation, while others mentioned measuring by the number of times AI directly exposes product or brand names in Q&A. Many provided "refund if no effect" and "measurable results" guarantees.

While GEO optimization is emerging, the industry ecosystem is still maturing. Notably, the technical leader responsible for AI internet search technologies like DeepSeek and Doubao expressed in a media interview that the company was identifying, screening, and downgrading low-quality AIGC content due to GEO chaos.

However, the suspicion that "response spaces" are turning into "ad spaces" is not unfounded. Where there is traffic, there are business opportunities. Toubao Research Institute estimates the domestic GEO market size at approximately 2.1 billion yuan this year, expected to reach 24.2 billion yuan by 2027, potentially reshaping over 300 billion yuan in market value over the next five years.

These scenarios inevitably remind one of "Black Mirror's" seventh season, where the human body becomes a "walking advertisement machine." Once keywords are recognized, ads interrupt the protagonist's consciousness, popping up uncontrollably at work or in love. While this may seem distant, as we rely heavily on AI responses, such a future could be closer than we think.