42 Zhangjing episode analysis · Product design and lasting experience
Guest: Fan Haoyu, senior vice president of Li Auto, as introduced at the time | Host: Qu Kai
Published: October 11, 2025 | Duration: 82 minutes 21 seconds | Prepared: October 18, 2026
Official episode and listening link
Why a useful feature may still fall short of a good product
Fan describes approximately 25 years in products and design across industrial design, phones, games, media, operating systems and cars. He identifies a common capability: observe, understand, combine and compress, turning numerous aspects of life into a coherent solution people can directly experience.
The official premise is a conversation about products and life, not disclosure of Li Auto’s business. Discussions of cities, generations, taste and AI’s future remain personal observations and hypotheses from 2025, not representative surveys. This analysis asks how they change product reviews while preserving the host’s challenges about taste being wrong or ahead of its market.
1. A self-righting fingerprint reader: Combine small details from real life
Episode reference: 13:56–21:28. Fan’s favorite early design was a fingerprint reader for a research office. Documents covered desks and could bury a small device. He made it a water-drop-shaped, self-righting object: shifting papers would reveal it through movement, while people could also play with it. Practical discovery and a small office pleasure shared one form.
This illustrates compression. Rather than listing ten features and satisfying them sequentially, find one solution addressing several conditions. Games, content and hardware can all begin with collecting material and connecting relationships before producing an answer. His concern about rushing to A/B tests is that weak hypotheses prematurely restrict alternatives, not that all experiments lack value.
He also describes a stimulation cycle: users adapt, suppliers escalate and both may become tired. This interpretation comes from conversations with creators, not universal causal research. Editorial inference: Alongside clicks and time spent, ask whether users finish tasks more easily, value the time they spent and experience extra burdens from continual changes.
2. Emotional value can mean different things to parents and children
Episode reference: 26:15–36:07. Qu uses parents breaking pizza down into flour and meat to illustrate practical thinking: why should it cost so much? Fan offers a counterexample. Older generations also entertain friends, buy good alcohol and cherish old objects or photographs. Emotional value is present but attached to different meanings.
This disagreement is more useful than labeling young people emotional and older people practical. Saving in everyday life and spending on special occasions need not conflict. One purchase cannot reveal a person’s entire value system. The same feature can mean different things at work, with family or alone.
Fan also uses opportunities, relationships and service availability in different cities as an observational framework. It reminds designers that their lives do not represent everyone, without establishing that big-city residents are inevitably lonely or smaller-city residents secure. A survey percentage retold without its report is not adopted as social fact. Editorial inference: Ask users to explain an actual decision, who was present and why it mattered, instead of assigning motives from age or city.
3. When taste and metrics disagree, make judgments specific and own the consequences
Episode reference: 39:42–47:46. Fan understands taste through distinction, style and content. A product becomes visible through continuing choices about experience, attitude and exclusions, beyond a feature list. Price and visual attractiveness are insufficient definitions.
When metrics conflict with feeling, he trusts experience gained through intensive use and contact with users. He argues that selecting indicators and categories already involves judgment. Like a practiced chef noticing an off flavor, a product designer may sense a version drifting before a dashboard shows an alarm.
Qu challenges this directly: taste can be mistaken or ahead of demand; a product good under one criterion may perform poorly commercially. Fan accepts responsibility for the result. Editorial inference: Turn discomfort into a specific hypothesis identifying the situation, people and damaged experience, then add direct feedback, failure examples and longer-term segmentation. Metrics cannot represent every experience, but intuition still requires scrutiny.
4. Wanting several things requires a practical foundation; software also resists change
Episode reference: 47:47–57:15. Fan grounds idealism in commercial survival. Cost, usefulness, technology, clarity and lasting meaning need simultaneous consideration. Design should seek better combinations, rather than unlimited features without resource choices. Asked to choose usefulness or interest, he would remove some activities and retain combinations offering both.
The host contrasts hard-to-change hardware with software. Fan points to software architecture and habits. Software resembles objects in someone’s home: moving them disrupts life; hardware resembles the house itself. Rollback does not make changes costless. Established entry points, records and routines make migration and explanation part of delivery.
For research, he discourages asking about technology terms directly. A sweet-potato seller may reject an AI assistant while wanting to sell more successfully. Begin with a falsifiable hypothesis, approach users and test the problem, solution and acceptance; revenue supports the next attempt. He likens the relationship to friendship, allowing disagreement and mutual understanding. That analogy cannot remove service promises or commercial responsibility.
5. Four “sounds”: Extend review from discovery to replacement
Episode reference: 59:54–01:02:00. This is Fan’s experience metaphor. Transcribed exclamations are unreliable, so the useful elements are four stages rather than invented slogans.
First, something stands out among similar products and invites approach. Second, initial use feels exciting and smooth; he gives installing apps on a new iPhone as an example. Third, months later, new capabilities still emerge, such as discovering a new application or activity. Fourth, on replacing the device, the old life, information and habits return smoothly. Delight may mean one fewer setup problem, rather than stronger stimulation.
Editorial inference: Collect evidence for discovery, first value, continued exploration and migration. Advertising clicks cover only part of discovery. A new-user demo cannot establish value months later. Without real migration, the fourth stage remains unproven. Claims that a certain number of sounds predicts sales are encouragement, not a forecasting formula.
6. Interest emerges through shared use, while AI expands responsibility for delivery
Episode reference: 01:06:22–01:20:18. Playing with lighters, writing on fans and adding laptop stickers show people transforming nearby objects into personal expression. Fan prefers discovering small life details and developing them with users to dividing interest into mechanical levels of emotional stimulation.
This can inform participatory design without guaranteeing meaning for everyone. Physical analogies and an unnamed psychology experiment are not treated as scientific evidence. Editorial inference: Observe what users add, share and sustain themselves, rather than first prescribing where they must feel moved. Users should also be able to reject an interaction.
The closing conversation returns to organization. Fan describes a design candidate using AI to demonstrate a complete tool, including account registration, rather than only icons or a requirements document. He expects AI to encourage responsibility across the entire path from need to outcome. That does not eliminate specialist collaboration or require one person to do everything. Defining problems, authorizing actions, handling failure and verifying results still require clear ownership.
Revisit a product across four stages
This is an editorial exercise, not Li Auto’s published evaluation. Recruit people at different stages rather than asking only newcomers to score the product:
- Ask a first-time participant what seems different and what they expect, then observe whether the first real task matches the promise.
- Ask someone with three months’ experience to demonstrate a recently discovered source of value, distinguishing capability from novelty.
- Ask someone who changed devices, accounts or versions to repeat a task, recording how information, habits and settings return.
- Rewrite “tasteful” or “hard to use” as a concrete episode, alongside cost, reliability and user explanations. Decide whether to preserve direction, change implementation or stop.
This suits continuing services, creative products and products embedded in everyday life. A one-off utility can have a narrower purpose. Interest need not be a purchase requirement, and the framework does not require adding personality to every tool.
Listening guide and related reading
- 13:56: The self-righting fingerprint reader.
- 18:20: Metrics and stimulation cycles.
- 32:26: Debate about generational practicality.
- 42:03: Taste conflicting with data.
- 53:43: Approaching real needs through hypotheses.
- 59:54: Reviewing four experience stages.
- 01:17:47: AI and complete delivery organizations.
Related themes: product methods, lasting experience and user research. Compare Albert’s evaluation and business framework and personal models understanding individuals.
Source boundary: Reorganized after reading the complete supplied transcript, prior analysis and official introduction, without sentence-by-sentence audio checking. Names and roles follow official material; unclear career details are not expanded. City distinctions, consumption and psychological explanations remain personal views. Future AI conversation proportions and changing roles are scenarios, not verified adoption. No specific Li Auto implementation or business outcome is established. Closing music is not treated as interview opinion.