GuShiio.com Wen| New Position Pro
Whenever the Internet surges, search is always the first target to be revolutionized. In 1998, Google used PageRank to attack Yahoo’s artificial catalog. Since then, the information hub of the PC era has been defined by search engines.As the highest-frequency entry point for user needs, search has naturally become the first testing ground for technology iteration.
The rise of social search has been frequently discussed in the past year. Media reported that Xiaohongshu’s average daily search volume in the fourth quarter of 2024 reached 600 million. In terms of the number of searches alone, Douyin, Fast Hand, and Station B, which are both content platforms and highly active users, also have active search ports, but Xiaohongshu is one of the platforms that most emphasizes its search functionality. In addition, the TikTok refugee incident at the beginning of the year added another boost to its traffic scale.
But the rapid changes in Internet competition are thatThe seemingly endgame may just be a preface.When Xiaohongshu gradually took over decision-based search scenarios with millions of real experience notes published every day, the variables brought by AI search exceeded expectations. Two years after its launch, ChatGPT has a monthly activity of 600 million yuan, and its search function was fully opened at the end of last year. Thanks to advanced language models and real-time web crawler capabilities, the AI search star product Perplexity performs impressive in terms of accuracy.
Traditional search is represented by Baidu and Google. The current advantage lies in real-time information indexing, and its functionality has long been tied to the ad-driven business model. Social searches, such as Xiaohongshu, use UGC content to build an information database and leverage the rule of traditional search with user usage time, but it is still limited to category coverage and commercialization pressure. AI search has made more technical rewrites, and DeepSeek’s popularity has also been driven by popularity, but it also faces multiple challenges such as answer interpretability, data compliance and profit model.
Open-wire is a transition in technology, and dark wire is the user’s choice.When technical routes begin to diverge, there may no longer be a unified search, only answer engines that adapt to different scenarios. While user needs are gradually shifting from information acquisition to decision assistance, the traditional search information mediation model also needs to be updated.
The interactive act of searching may be transcending where to look, but who defines the answer.
01 Fast forward and fast out and immersion consumption
If we want to know news related to the Spring Festival movies in 2025, enter these words into the search engine, and the movie introduction, box office records, and latest film reviews will be aggregated on the return page in real time. This is the most solid barrier for traditional search engines in the current changing market situation.Comprehensive and timeliness, especially in scenes such as breaking news, stock quotes, and live broadcast of events.
Distributed crawler and hot indexing technology allow search engines to monitor and crawl news sources in real time. However, behind this moat, its content ecosystem is difficult to compete with UGC-style content platforms. “New Position” mentioned in previous articles analyzing search and marketing thatWith the rise of social media and content platforms, the way users obtain information has undergone fundamental changes.
According to QuestMoblie data, for typical new media platforms (Douyin, Fast Hand, Xiaohongshu, Beili, Weibo), the number of de-heavy active users has reached 1.071 billion in October 2024, with a penetration rate of 85.7%. In terms of single-day usage time, Xiaohongshu has basically exceeded one hour, and Douyin Fast Hand is on the order of nearly two hours.
Both traditional search and social search have rich information sources, and the more critical differentiation lies in user behavior and the duration of stay it affects.
For example, young people nowadays prefer situational expressions, adding more emotional symbols or practical details based on keyword matching. For example, compared with women’s clothing, the phrase more commonly used by young people when commuting clothes look slimmer, restaurant recommendations can be refined into scenes such as cooking tea around a stove, or the concept of eating and drinking by special forces nbsp;
The intergenerational fault in behavior has produced more long-tail keywords, which are somewhat misaligned with the target index logic of traditional search engines, but are just covered by the UGC output of content platforms (such as the real experience of Xiaohongshu Notes). nbsp;
Moreover, the information returned by social search is closer to the decision-making link in the consumer chain. Traditional search scenarios are more fast-in and fast-out news viewing. After the peak of breaking news traffic, users quickly leave. Advertisers are becoming increasingly reluctant to pay high prices for instant attention, and commercial value has shifted as a result. nbsp;
Higher monetization efficiency attracts PUGC output, content libraries enrich and amplify the information advantages of built-in search, and awareness of search positions shifting to content platforms continues to spread, but not without obstacles.Social search also faces the problem of commercial value not eroding content quality.
For example, Douyin’s 2022 test concluded that once the e-commerce content displayed exceeds 8%, the user retention and user usage time of the main station will be significantly negative affected. Whoever can turn answers into commercial value without destroying trust has the initiative.& nbsp;
Another variable that is more real-time and more global is that AI may rewrite the rules of search.For example, some AI search tools have launched real-time networking capabilities, covering some timeliness requirements, and the advantageous window period of traditional search has gradually shortened. The potential disruptive effect of technology has made the two search forms mentioned above incorporate the logic of AI into change considerations. nbsp;
02 AI search, replace or supplement?
Data from market analysis firm Gartner predicted in February last year that by 2026, search volume by traditional search engines would drop by 25%, and their market share would be lost to AI chatbots and other virtual assistants. nbsp;
As of now,”New Position” believes that applications involving AI search concepts on the market are roughly divided into two groups,Independent applications and embedded tools,Of course, there are also companies that bet on both sides.& nbsp;
Independent applications use AI as their core selling point and aim to provide a completely reconstructed search method from the bottom to the interactive experience. Typical domestic products include Kimi, bean bags, secret towers, etc. Most products have dialogue memory and source indexing functions. Depending on different positioning, some focus more on providing multi-scene agents to play ChatBot, while some focus on professionalism and are committed to improving model capabilities. nbsp;
For regular search needs, these applications and traditional search engines still have their own advantages and disadvantages. The biggest feature of AI is that it shortens the entire information chain and directly presents structured natural language, which reduces the time cost and cognitive burden of obtaining information to a certain extent.But if the user wants to obscure the search, or has the need to evaluate the source, this is still the home of traditional search engines.& nbsp;
The value of independent applications lies more in functions that are not traditionally available, such as tabulation, long text processing, and personalized interaction. But we still believe thatFirst, because independent applications require users to change their search habits, user migration costs are relatively high. Second, the non-interference answers provided by AI search are not compatible with the advertising bidding ranking business model, and the prospects for realization are not yet clear. nbsp;
Embedded tools are well understood, that is, integrating AI functions into existing search engines or platforms. It follows the original portal and relies on the ready-made search ecosystem to lower user education costs, and the platform can also call endogenous data to optimize the generation effect. nbsp;
Embedding AI has indeed become a fashionable proposal for product upgrades in the past two years. Alipay launched Zhi Xiaobao and 360 launched nano-AI search. The Xiaohongshu mentioned above is a case of betting on both sides. After successively launching in-station AI search assistants Sousou and Da Vinci, it launched the independent AI search App to the front desk at the end of last year. nbsp;
However, these businesses are all rooted in the original content ecosystem, and the purpose is still to use AI to find increments. As mentioned in Diandian’s promotion, by aggregating life experience across the Internet, users can find and summarize answers to related questions such as food, shopping, and travel. nbsp;
“The advantages and disadvantages of supplementing or transforming ideas are that they have not completely innovated the search method.The advantage is that it can still rely on the business model of traffic monetization, allowing users to get some icing on the cake experience given by AI; the disadvantage is that due to user adaptability and resource investment considerations, technical iteration under this idea may be relatively conservative. nbsp;
It is undeniable that the upper limit of AI is still breaking through our imagination, just like the cognitive impact DeepSeek brought to everyone at the beginning of the year. butCurrently, in the field of search, smooth transition is still the mainstream. It gives traditional search and social search the same opportunities to overtake with the help of technology. The difference lies in who can better translate the needs of the market and users.
03 Written at the end
At present, the three search forces are in a commercial intermediate state: the old model is failing and new rules are not established. nbsp;
Against the background of shortening the window period for timeliness advantages, traditional search engines need to be transformed most; social search must balance content ecology and advertising density to stably meet the needs of a wider range of users; and independent AI search applications are still at high cost. At this stage, it remains to be seen whether existing subscription systems or API charges can support large-scale profits. nbsp;
The influx of users will accelerate this series of explorations.As of 2024, 230 million people in China have used generative artificial intelligence products, accounting for 16.4% of the total number of Internet users. At the beginning of 2025, DeepSeek will be the driving force for the popularization of AI tools. Usage habits shape product forms, and interactive feedback promotes technology upgrades. If in the last round of search revolution, users found a more emotional side of search on social platforms, for AI, a rational product, the final finalization still needs to refer to a lot of user feedback. nbsp;
But we can at least establish a consensus.The ultimate form of search must be the resonance of business logic and technical route, and only the first player to find the resonance frequency can define the next decade.& nbsp;
