Has user search behavior changed? From "keywords" to "conversational questions"
Chapter 1: Awareness · Question 6
A massive shift is underway
Let me ask you a question first: has your own search habit changed recently?
Think back: are you less and less likely to type something like "Beijing hotpot recommendation" into Baidu as aKeywordcombination, and more used to opening Doubao or Kimi and asking—“Are there any hotpot restaurants near Beijing Guomao suitable for a friends' gathering, around ¥150 per person?”
This is not an isolated case.User search behavior is shifting comprehensively from "keyword assembly" to "conversational questions"—the biggest paradigm migration in search habits over the past two decades.
The essential difference between the two search modes
The keyword era (mainstream 2000-2023)
Users were trained to be "translators"—translating their needs into keywords that search engines could understand:
- Want to find acne products → type "acne treatment effective student"
- Want to learn about investing → type "2024 investing stable recommendation"
- Want to buy running shoes → type "running shoes men cushioning under 500 yuan"
The conversational era (accelerating since 2024)
Users are starting to express real, complex needs in natural language:
- “I keep getting recurring acne on my face, mainly on my chin, possibly related to staying up late—are there any affordable skincare solutions?”
- “I have 50,000 yuan in spare cash, don't want to take on too much risk, but feel it's a loss to just leave it in the bank—any suggestions?”⟧
- “My knees aren't great and I want to start running for exercise—what shoes with good cushioning can I buy with a 500 yuan budget?”
Comparison at a glance
| Feature | Keyword search | Conversational questions |
|---|---|---|
| Expression style | Fragmented phrases | Complete natural sentences |
| Information density | Low (3-5 words) | High (includes scenarios, constraints, personal circumstances) |
| User expectations | A bunch of links, filter them yourself | Directly give personalized recommendations |
| Typical platforms | Baidu, Google | Doubao, Kimi, Xiaohongshu search, Douyin search |
| Requirements for content creators | Keyword stuffing | Truly solving problems |
Data evidence: the change has already happened
This trend is especially pronounced in the Chinese market:
- Xiaohongshu (RED): search volume has surpassed Baidu in some categories, and the average length of users' search queries continues to grow, increasingly resembling "questions" rather than "keywords"
- Douyin search: daily average search volume has exceeded hundreds of millions of times, with a large number of searches beginning with interrogative words such as "how"“"why"“"recommend"
- AI assistants: Doubao's monthly active users have exceeded 300 million (source:QuestMobile"2026 H1 China Mobile Internet Strength and Value Ranking"), and Kimi's long-text conversation volume is growing rapidly—users have already become accustomed to "chatting with AI" to get information
- Baidu itself: is also adding AI-generated summary answers to its search results, showing that even traditional search engines are adapting to the conversational trend
Note: The search volume and user behavior changes of the above platforms are based on public reports and industry observationscomprehensive estimates, not official data disclosed by each platform; for AI assistant MAU scale, please refer to QuestMobile public rankings.
What does this mean for brand marketing?
1. Content can no longer be written only for "keywords"
The era of ranking by stuffing keywords like "acne product recommendations"“"skincare for students" is coming to an end. AI systems understand semantics and don't buy into keyword density.
2. You need to answer "complete questions”
Users ask complex questions with context, and your content also needs to provide complete answers with context. Fragmented product introduction pages are increasingly difficult to get recommended.
3. Long-tail demand becomes a new opportunity
Conversational questions naturally contain more qualifying conditions (budget, scenario, personal situation), which means more segmented long-tail demands are surfacing. Small and medium-sized brands can precisely position themselves in these specific scenarios.
4. The logic of “being found" has changed
In the past it was "user searches keywords → sees your link → clicks in”. Now it's "user asks a question → AI directly cites your content as the answer”.You no longer need to be clicked; you need to be cited.
Practical recommendations
- Content topic transformation: Stop planning content around keyword lists, and instead collect users' comments in Xiaohongshu comment sections,ZhihuCreate content centered on the real, complete questions from inquiries and customer service conversations
- Simulated AI Query Test：Regularly enter conversational questions your target users might ask into tools like Doubao and Kimi, and observe what answers the AI gives and whose content it cites—if it's not yours, analyze the gap and fill it in accordingly
- Upgrading content structure：Each piece of content opens by directly answering a specific question, and the body offers scenario-based, condition-qualified advice (rather than generic statements), making it easier for AI to extract as a precise recommendation
