Code Engraved | How Li Auto Built a Miracle: ¥100B Revenue, ¥10B Profit

On February 26, MaHui member Li Auto (NASDAQ: LI) released its Q4 and full-year 2023 financial results, achieving the best financial performance in the company's history.

In 2023, Li Auto delivered 376,000 vehicles, up 182.2% year-over-year; revenue reached 123.85 billion yuan, up 173.5% year-over-year; net profit hit 11.81 billion yuan. In the fourth quarter just past, Li Auto captured 16% market share in the above-200,000-yuan new energy vehicle segment, ranking first among Chinese automotive brands.

It took Li Auto eight years to achieve 100-billion-yuan revenue and 10-billion-yuan profit, making it the first among China's "new forces" in car manufacturing to achieve full-year profitability — a miracle in the growth of Chinese automotive brands. Source Code Capital, as Li Auto's earliest institutional investor, was a major participant in the company's Series A round and continued to increase its investment across multiple subsequent rounds, with cumulative shareholding ranking among the top two financial investors. At the founding of Source Code Capital, Xiang Li, as a standout entrepreneur, became a limited partner in Source Code Capital's fund; at the founding of Li Auto, Source Code Capital in turn became Li Auto's Series A investor. Xiang Li once said: "Source Code Capital built MaHui as a healthy and sustainable industry connector, providing efficient learning and exchange opportunities with China's top internet companies and founders. Source Code played a major role in Li Auto bringing in investors like Meituan and ByteDance." Last year, Li Auto Chairman and CEO Xiang Li publicly shared his product methodology. From exploring a 100-billion-yuan market to surpassing user needs, Xiang Li and Tuo Buhua shared their insights from over 20 years on product, entrepreneurship, organizational management, and continuous growth. Source Code Capital is honored to have accompanied Li Auto for eight years, redefining the product of "car" and the experience of "driving, riding, and using," delivering an outstanding report card through delivery volume and user reputation, and validating Source Code Capital's continued ability to generate excess investment returns. Source Code persists in investing in technology-driven innovation and in the creation of lasting, real value. We share this article again today, to win new momentum together with all of you in the new year! Article from: Luoji Siwei WeChat account (luojisw)

Original title: Tuo Buhua in Conversation with Xiang Li: How to Turn Luck into Capability

1

How to Position Yourself to Explore a

"100-Billion-Yuan Market"?

Tuo Buhua: What we're hearing today is the making of a 100-billion-yuan blockbuster product. It has already been tested by a massive market. At the same time, this market is also highly mature — other markets are called red oceans, but yours might be called a "bloody ocean," right? Just before the livestream started, Xiang Li told me that making cars is incredibly hard, incredibly draining. So I asked him, is this drain mainly internal or external? Could you explain to everyone what's so special about the auto industry?

Xiang Li: I think intelligent EVs have a unique characteristic. If you look at Toyota or Volkswagen, they grew up alongside the entire automotive industry. But when it comes to intelligent EVs, ICE vehicles were still selling, and traditional automakers were extremely dominant. If you look at the United States or Europe, over the past 30-plus years, no new company has survived besides Tesla — which proves that the barriers to entry in this market are extraordinarily high.

At the same time, cars involve a long value chain. It's a bucket theory: if you have production capacity for 300,000 vehicles but can only sell 100,000, then your 300,000-unit production capacity is wasted — that's internal friction. If you have capacity for 100,000 but can sell 300,000, you can only deliver 100,000, and your sales infrastructure becomes massive internal friction. So I see this as a complex, long-chain industry, but we've made it even heavier. Why? Because we're pursuing intelligence. In the past, intelligence could be handled by suppliers. Today we have to do it ourselves — we're taking on suppliers' work. So we're fundamentally an automaker, but we also have to be an internet company. Even our sales systems are built in-house. We also have to be an AI company and a traditional supplier. Moreover, while legacy automakers simply manufacture cars and sell them to dealers, we've chosen direct sales — so we're also doing what automotive 4S dealership groups do. The requirements are completely different now.

At the same time, because several of us went public in the capital markets from the very beginning, whether from the media's perspective or investors' perspective, people examine us under a magnifying glass. They hold us to Mercedes-Benz, BMW, and Audi standards. You have to deliver fundamental vehicle quality, but also excel in intelligence and business model — only then do they believe you have a chance to survive. In such a complex industry, you must maintain an entrepreneur's mindset, considering every detail thoroughly from the early stages. This is incredibly draining on people. So when the three of us from Wei Xiaoli — NIO, Xpeng Motors, and Li Auto — get together, every time we meet we notice how rapidly each other is aging.

Tuo Buhua: But you truly are a young veteran. Whether in entrepreneurship or in automotive, Autohome gave you deep familiarity with the entire industry. But I honestly hadn't thought about this before you mentioned it — these demands are somewhat unreasonable for you.

Of course, I think any kind of entrepreneurship is remarkable. But if I were starting a milk tea shop, when I open that first location, I only need capabilities matching that one shop. Doing that one shop well would already be great. But for you, it's equivalent to needing the capabilities to operate 10,000 locations before you can open your first shop and sell your first car. The resource matching, including timing — it's extraordinarily difficult.

Xiang Li: Fortunately this is my third time founding a company. I was mentally prepared, because during the entire mobile internet era, Autohome only achieved a vertical category — essentially a second-tier business model within mobile internet overall. Looking further ahead, from the perspective of the entire AI field, I definitely wanted to choose a mainstream, most difficult path. No matter how hard, I had to pursue the mainstream.

Tuo Buhua: Actually this is also the first formal question I especially wanted to ask you today: How do you position yourself to discover such a 100-billion-yuan market? Because small money comes from effort, big money comes from choice. Most of the time we spend our energy on small efforts, but when it comes to selecting tracks, markets, and users, there seems to be little experience to draw upon. For you, how did you choose this track?

Xiang Li: I think this is a fundamental difference between what we've done over the years and most other new EV forces.

I look at it from two dimensions. The first dimension is technology, because creating an entirely new massive industry from scratch is basically very difficult. But I think there's opportunity in a massive industry undergoing huge technological change, where you can become a new leader. The second dimension is users — how we can effectively capture user groups.

Looking at technology first: for humanity as a whole, the two aspects where we can ourselves drive technological impact are energy and information. So why did we choose intelligent EVs? Because intelligent EVs cover both. Intelligent EVs adopt new energy, and energy methods will continue evolving.

Intelligent EVs differ significantly from traditional cars in that they carry their own critical energy battery pack, whereas traditional cars convert fuel into driving force through internal combustion and that's it. So if we want a platform capable of computing, electrical energy is critical — I think this is a huge benefit that new energy brings.

The second major benefit is AI. Our internal view is that AI can simultaneously transform both the physical world and the digital world. So finding the intersection point that changes both the digital and physical worlds is the intelligent EV, because it spans both. We even believe that this mother ecosystem spanning physical and digital worlds will emerge from intelligent EVs. So when you mentioned Li Tongxue earlier, including our assisted driving — Li Tongxue represents our construction of the digital world, while autonomous driving and intelligent driving represent our AI transformation of the physical world.

Tuo Buhua: When Dr. Wu Jun discusses tech history, one of his conclusions is that the progress of all human technological history is driven by two forces: one called energy, one called information — just these two things.

Xiang Li: So recently room-temperature superconductors became hot, because there haven't been many fundamental breakthroughs in the underlying technology of the energy field for years — mainly optimization.

Actually, what we've been exploring over the past eight years is genuinely changing. The new energy vehicles we've made are genuinely better than what traditional automakers produce. The AI technologies we've developed are far beyond their reach. So while they're incredibly powerful — each company possibly generating over a trillion in revenue — they still cannot stop our development.

Tuo Buhua: Could I understand it this way: In your process of selecting this 100-billion-yuan market, you're essentially seizing an upgrade or renewal opportunity in an existing market?

Xiang Li: My entry point is here. But what we'll create in the future — I think as I just said, intelligent EVs enable AI to span both digital and physical realms. It may be a mother ecosystem. Just as Apple, in many ways, created the iPod before the iPhone came after, where you choose to enter is crucial.

The other aspect is user needs. As a startup, to survive in an industry, I must enter a high-growth sector. Regardless of market size, I cannot enter a shrinking market.

I don't claim we're some evangelist for intelligent EV普及. Tesla, founded 12 years before us, is the world's most valuable company. BYD is China's most valuable automaker. They're both incredibly powerful, and their founders are working desperately to run their companies.

How do we look at entry points? This certainly relates to my experience with Autohome. We looked at where the major growth opportunities lie, and ultimately we chose to serve family users above 200,000 yuan.

People might hear "family users above 200,000 yuan" and think that's a narrow market. Actually not. By the numbers, family users above 200,000 yuan represent 81% of all car buyers above 200,000 yuan — so this is a massive market. And this market is still growing rapidly. Why? Although China is a large country, urbanization is concentrating toward major cities. Road resources, parking resources are extremely limited. So what phenomenon emerges? Buying a car — with Beijing's license plate restrictions, Shanghai's license plate restrictions — becomes unfair for many young people. They rent apartments without parking spaces. They work at companies where parking is distant and expensive. So today, ride-sharing is more economical; buying a car is very difficult for these users.

So we can see that cars below 200,000 yuan, especially below 100,000 yuan, their buyer demographic is declining significantly, because sub-100,000 is generally first-time purchases. But family user purchases above 200,000 yuan are growing continuously.

Why? Because I need to drive my child to school in the morning, pick them up in the evening, take parents and children out on weekends — this is something ride-hailing cannot replace, something sharing cannot replace. At this point, buying a car becomes a necessity. And because parking is expensive, because license plates are hard to obtain, when people buy or upgrade, they tend toward upgrade consumption.

Tuo Buhua: Buy something better.

Xiang Li: Right, so my market must be a growing market.

Second, in this market we must be able to win. We must ensure we can win — this need isn't being met by others.

In the past, no matter how much I spent on a car, even over a million for an SUV, the experience for my children, for my parents, was terrible. Because all cars were designed around the front-row driver and the rear-seat boss. For my spouse, for parents, for children — a child sitting in the second row of a seven-seat SUV is like behind a city wall, like being locked in a cage. Elderly people getting in and out, adjusting seats — all extremely inconvenient. The entire experience was terrible.

We wondered: How can we build a great car for a family user — the husband, the wife, children of different ages, and our parents? When everyone in the company is focused on transforming these things, such products and experiences are quickly perceived and accepted by users.

Tuo Buhua: Here I must give feedback as a car owner. Before buying the Li Auto L9, I had driven cars from about eight brands. I truly upgraded to a larger car because I had children.

So basically from when prenatal checkups became regular, switching cars became a necessity. Safety became priority one. Size became priority one. Whether a pregnant woman could get in became priority one. And three generations traveling together became a must-consider factor.

But before driving the Li Auto, I had driven large cars from two other brands. I felt that although the car looked bigger, the problem wasn't solved.

For example, one obvious issue: you definitely need a third row, otherwise it's not enough for family outings. But whoever went to the third row was unhappy, because first, the third row was uncomfortable to sit in; second, the third row was hard to get into, you had to crawl; third, the third row was very enclosed.

At the time I thought, fine, let the child sit there — so small, just strap in the safety seat and sit in back. But actually children didn't want to either, because it's so enclosed, they can't interact with others, it's very cramped, and air circulation is poor.

But after I switched to the L9, I felt this problem was solved. I found my child would voluntarily go sit in the third row, and her experience didn't change at all. I think this is truly remarkable.

Xiang Li: Right, it's all in the details. It's really about standing genuinely in the user's shoes and thinking about value. So if you look at the Li L9, the third row is actually higher than the second row. The front seats don't block the kids' view — it's like a movie theater. So whether they're sitting in the third row for conversation or for the view, their experience isn't compromised at all. The big side windows, the sunroof overhead — all of it was designed for the children.

How do you find target users through the lens of "life stages"?

Tuobuhua: In the second lecture of our course, you went through a detailed analysis of how you chose family users. One tool really stuck with me. Most marketing and management books talk about user selection, but the vast majority use statistical or demographic methods. You invented a different approach — finding positioning through life stages. I found that mind-opening, because it's not simply dividing by income, by city tier, or by age bracket. It's starting from when someone enters that phase of life where they need to get the whole family out the door together. How did you discover this tool or this logic?

Xiang Li: Actually, the core thing for us was seeing that the needs of people around us weren't being met. No matter how wealthy my friends were, this kind of need simply wasn't satisfied. So we dug deep into that scenario. A lot of the time it's just thinking one layer deeper. "Family user" isn't enough — which stage of family? A family of three? Three generations under one roof? A couple with kids and a nanny? We looked at scenarios like: is the child under ten or over ten? Completely different use cases. I think that's the opportunity we spotted.

Tuobuhua: Right. There's this old idea that to build products you have to understand people. But there are different depths to that understanding. From Li Auto's experience, we can see that understanding people demographically is one thing, sociologically is another. What you've done — digging into psychology, demographics, sociology combined — is returning to the user's real state. That's yet another level. For those of us building products, I think their approach to defining cars and excavating user needs through life stages is an excellent lesson.

This wasn't the first time Li Auto used this user-definition method. In the course I also saw a case about Autohome — you defined Autohome's users as car consumers, not car enthusiasts. Could you expand on that?

Xiang Li: I think for any company, it's crucial to ask: why do positioning at all? It's to answer who we are, who we serve, and what value we create. It's about internal efficiency and external communication efficiency both. A good test is: when I describe a user group, the best outcome is I ask ten people and all ten understand it the same way. So back when we were building Autohome, we debated this definition internally for a long time: are we serving car users or car enthusiasts? We finally decided on "car consumers" — if he's buying a car, we serve him; if he's not buying a car, we don't serve him.

Because the more information I provide, the more it distracts the user, and the lower the value becomes. So everything you saw on Autohome — at least everything we provided while I was there — was about helping people buy cars and use cars well. Traditional motorsports, F1, WRC rally racing — we didn't touch any of that.

Tuobuhua: That stuff is pretty cool though, and it gets traffic.

Xiang Li: Right. And many sites did car models — we didn't do that either, because that's for people who don't necessarily buy cars but just like cars. There are specialty sites for that. Another important thing: if you went to any auto show, there'd be all these model girls. Autohome was the only one that didn't photograph them. If you want to look at pretty women, why come to us? We were the only one that stayed focused purely on cars. This meant the user base we attracted were people actually buying cars, so our commercial value was higher too — for all our users, the ads we ran, there was no invalid traffic, no fake traffic.

Tuobuhua: That's remarkable. As they say, small money comes from effort; big money comes from choices. What is choice or decision-making ability? I think this is the most important capability for a founder — choosing between two right options, that's the hard part. The racing events, the car models — from a traffic perspective they had traffic, from an ad perspective they had ads. But why choose not to do them? Being able to answer that clearly behind the scenes — I think only then can a product truly be done well.

There's one detail that particularly moved me. Autohome's way of describing things was very distinctive. For example, describing a dimension — they wouldn't just say it's 23 centimeters. For most people that's hard to instantly grasp. But they'd say it's about as long as a water bottle. For someone like me who was buying a car, I immediately understood the size.

Xiang Li: We described it like this: fists and fingers. Legroom was how many fists, headroom was how many fingers. Because that's how people actually measured things — I'd sit in a Mercedes-Benz, which was basically the benchmark, and note the headroom in fingers. Then I'd sit in a BMW and check the headroom in fingers. And for storage space in the car, people would ask about cup holder size — users had no concept of that. So we'd say you can fit a Coca-Cola here, and a Mizone there. That's how we described space.

Tuobuhua: Today if you buy clothes on most websites, they'll tell you the model's height and measurements so you can compare — it makes decision-making so much easier. Autohome was doing this method many, many years ago. I think that's remarkable insight. So if anyone wants to position themselves in a hundred-billion-scale market, or define a mainstream large track, I think the methods Li Auto has explored and the tools they've summarized will definitely be very useful for everyone.

---

How do you surpass user needs in a red ocean market?

Tuobuhua: We just said the auto market, especially the smart auto market, isn't a red ocean — it's a blood ocean, where people stop at nothing. Consumers are being pampered extremely well because every company is trying hard. One day I saw a photo a friend posted — a huge traditional automaker's 4S dealership. He went for maintenance and had a luxury meal at the 4S shop. I haven't been to a 4S shop in so long, so I was shocked — has fawning over consumers reached this level?

In your process of building cars, you proposed a concept called "surpassing user needs." I think this is extremely difficult. Because discovering user needs is already hard enough. In this red ocean market, it seems like most user needs are already pretty well covered. Meeting them itself demands high capability. But you mentioned surpassing them. Could you expand on what surpassing user needs means to you?

Xiang Li: I think it's based on understanding long-term user value — though this is also related to our industry. The automotive cycle is very long. Product development takes 24 to 36 months. After that, the product stays on market for 3-6 years. So you're looking at a six-year starting point at minimum, and you can't make major mistakes in between.

Tuobuhua: You can't be careless.

Xiang Li: Right. If you simply satisfy user needs, I think it's easy — at least in this industry — to become "treating the head when it aches, treating the foot when it hurts." What's the long-term view? Take our vehicles as an example. We realized that for users to switch from fuel vehicles at scale, one crucial thing is solving their long-distance charging problem. Most homes can install charging piles, but you also have to solve the cost problem. Then we saw that range extension — though this technology existed many years ago — could be upgraded to solve problems for the next 10, 15 years.

Tuobuhua: I have to press on this point, because I remember very clearly — when you first announced the range-extender solution, the controversy was enormous. Many people said this was backward technology, that you'd chosen a backward technology from the start and were on the wrong path. But for you, if I recall correctly, you argued with many people many times. Why were you so firm?

Xiang Li: I don't think we cared that much about ourselves. We cared about long-term user value. Because for users like yourself, switching from a fuel vehicle to an EV — and with children at home — you can't say that because I switched to an EV, I can no longer take long-distance trips. You can't say that because I switched to an EV, if something comes up in Jinan, I suddenly can't go. For China's 200-million-plus vehicle fleet, for the over one billion passenger vehicles worldwide, this is a major problem. So our core focus was how to solve this problem. Because you also know the benefits of EVs — smooth driving, safety, quietness, low cost — but I have to solve this problem first before users can enjoy those benefits. Otherwise, if I buy an EV and many functions I had before just disappear, and these were the very meaning that cars brought to life — I think that's unacceptable.

So we looked at this problem and upgraded the traditional range-extender technology, turning it into a large battery. In the past, range extenders or PHEVs could only go thirty or forty kilometers. We achieved 200 kilometers, enabling fully electric driving in the city. But for long distances, you could still use China's gas station resources on the highways. This also avoided the situation where, after finally getting the whole family out for a weekend or holiday trip, you'd have to queue for two or three hours to charge — that's not a trip anymore, that's going out to be frustrated. So this is what we did based on what we saw.

At the end of 2015, beginning of 2016, when we were greenlighting this — if we had done user research and asked whether they needed range extension, users definitely would have said no. And the L9 you bought is a six-seat SUV. You should know that in 2016, there were no six-seat SUVs. SUVs were all seven-seaters. And as you mentioned, you'd have to climb to the third row, and third-row passengers faced a wall. So at that time we realized — how could we make it convenient for third-row passengers to get in and out without interfering with second-row passengers? In a traditional seven-seat SUV, if my parents sat in the second row, they'd have to get out first. I'd have to get out from the driver's seat, flip the seat forward — because it was all mechanical — the child would get out, then I'd flip the seat back.

Tuobuhua: Lining up to board.

Xiang Li: Right, the experience was terrible. In the end, seven-seat SUVs would have the third row folded down and be used as five-seaters. We wondered if we could enable direct walk-through, with elderly passengers able to adjust seats electrically — that would be so convenient. So we combined MPV functionality into an SUV. But if you'd asked users at the time whether they needed a six-seat SUV, they'd definitely have said no.

Tuobuhua: Users definitely would have said, I want seven seats, feeling like six seats is a rip-off.

Xiang Li: Right, they'd feel like spending the same money for one fewer seat.

Tuobuhua: Exactly, it feels like getting shortchanged on six seats.

Xiang Li: Or they'd say, I want a big trunk, so I'll just get a five-seater. But you know from experience that once you install a child safety seat in a five-seater, you can only fit one other person in the back — it's basically useless. So we saw that scenario and built a product for it. But if you'd asked users at the time, they'd definitely have said they didn't need it. Same with our screens and array microphones, same with Li Tongxue [Li Auto's voice assistant]. If you asked about voice control, they'd say give me more buttons, I don't want so many screens. Today you can see more and more companies doing range extenders — almost every single one is doing range extenders now. And almost every three-row SUV from every company is a six-seater. Even something like 50% of new cars now have the same screen layout as the Li ONE, though our L series has taken it further.

I think this is what it means to go beyond what users ask for. When you actually put it in front of them, they say this is exactly what I wanted — they just couldn't express it before.

Tuobuhua: This is what Steve Jobs meant when he said users never actually knew they needed an iPhone. It's a peculiar thing. I won't be embarrassed to tell you — when I was ordering your car, my whole concept was that kids shouldn't be looking at screens for that long. So I actually asked your sales rep if they could remove the rear screen for me. Like, no cartoons in the car for my kid. And your sales rep very bluntly told us, it's all standard equipment, can't remove it. At the time I made the choice reluctantly. Later I realized thank god we had that screen. Long drives, traffic jams, even my own relationship with my kid while I'm driving — none of the negative scenarios I predicted actually happened, but my kid's experience got so much better.

Xiang Li: We thought through a lot of details. Our screens filter blue light far better than blue-light glasses. And when natural light can come in from outside, there's actually no problem for children's eyes. The core cause of myopia in kids is getting too little exposure to normal outdoor sunlight, spending too much time indoors — we considered all of this. I have kids myself. I would never ruin my own children's eyes.

---

How to uncover user needs?

Tuobuhua: If these are needs users would never bring up even if you asked them, then when I'm building products today, how can I actually hit these needs? Or it's not even "hit" — they never mentioned them, so how do I dig them out? What methods do you use?

Xiang Li: I think the first important thing is, when we hire product managers, they need to be very sensitive. Because a lot of people are especially good at making do, and people who can make do aren't suited to be product managers. Someone who likes to complain, but whose complaints are justified — who's very sensitive — I think that kind of person is actually suited for product management. Give them opportunity and they can fix these problems, rather than feeling indifferent and letting them pass. So I'd say being especially sensitive to experience, that's the first thing.

The second especially important thing is that they dare to stick to their own views. A lot of times their view is right — it's just different from most people's. And in the process of sticking to that view, when everyone pushes back and forth, you often end up exploring toward the truth.

The third especially important thing is having the ability to rationally validate — to verify through various means whether this scenario actually exists. Honestly, building big products versus small features is different. The capability requirements for someone building big products versus small features are significantly different.

Tuobuhua: That's quite counterintuitive. Normally when we hire team members we want someone who fits in really well, but maybe fitting in well is precisely because they can make do — which might make them unsuited for product management. The second is being able to persist.

Xiang Li: Persist in your own views.

Tuobuhua: Does this persistence turn into conflict in your team? Like because they're so persistent, completely different from everyone else.

Xiang Li: That's why the third thing is validation.

Tuobuhua: Need the ability for rational validation.

Xiang Li: Right, to take something intuitive and make it rationally verifiable.

Tuobuhua: After you have this kind of team, what method for uncovering user needs would you most recommend?

Xiang Li: I think it's still about establishing a shared language for everyone, a common language that lets people communicate together. We call it PEA — Product Experience Analysis.

This is something we built when we first started, very much like AI. First it needs broad insight into the entire problem space that needs solving, and this insight doesn't just come from the automotive sector. It might look at smartphones, even drones, even some sociology — so that its overall input is strong enough.

With sufficient input, and it's created together — for instance, product doesn't do it alone. It pulls in technology, might pull in marketing people to do this insight together.

The second step is establishing the target for the problem to be solved — what exactly am I solving for, and the corresponding plan, to make sure everyone sees the same thing and is building toward the same thing.

The third is actually executing and operating, and execution and operations aren't the end either. This includes having the ability to assess costs during execution and operations. Otherwise people are just shooting from the hip building a bunch of zombie products. You need to assess costs, and assess operational results.

So a lot of times we say why don't we hire product managers straight out of school? A product manager fundamentally needs to be an operator.

Tuobuhua: Very comprehensive capabilities, huge span.

Xiang Li: Right, they're not just doing product. They need operational awareness and experience behind the product, and basic financial capability too, so they can be rational.

The last thing is actually after finishing, how do we review and distill that into a capability. Very much like how people build AI today — input, model, output, and training. In this approach, and because everyone works this way, they can see others' best practices, and others can coach them. That brings huge benefits.

Tuobuhua: PEA should be a working method you invented.

Xiang Li: Right, before the big mainstream rollout, all the big demand analysis, including working with Dedao to build this course — our team did a PEA demand analysis for that too.

Tuobuhua: Right, I was quite shocked. At first when a colleague told me you were willing to come share this course on Dedao, I didn't really believe it. I thought you might have acted on impulse, because you're too busy. Like you said, the auto industry really consumes people. I said there's no way you'll finish, being a Dedao instructor isn't that easy.

But even when our product was finally about to launch, I still wondered where you found the time for this? And where did you find the motivation? So I asked you earlier, why do this? You said it's because your team did a PEA, and the team also needs to master this product demand analysis working language.

Xiang Li: Everyone needs to master it, because other departments need to coordinate too. For instance, if they're doing an auto show, doing demand analysis for this year's show, it includes what sales needs to do, what product managers need to follow up on. It's all like this — it gets everyone aligned.

Tuobuhua: Right, it's like you've created a unique "Li Auto language" inside the company, translating everyone's various working languages from their past experiences into a standard language.

Xiang Li: Right, our collaboration on a course is also a product. Our doing an auto show is also a product.

---

How do you build an "efficiently collaborative" all-star team?

Tuobuhua: So everything is a product. When we got to this point, we actually touched on something I was especially curious about. Honestly, Dedao is only a few hundred people, and I already find managing this team exhausting. You're at twenty thousand.

Xiang Li: Over twenty thousand.

Tuobuhua: And it'll be more, because that's the inevitable scale of the auto industry. You also have a very famous line: treat the organization as a product too, build the organization like you build a product. I especially want to ask — how do you build a twenty-thousand-person team as a product? In this process, what do you think is most worth sharing?

Xiang Li: I'll talk about two relatively important phases, to help everyone understand better.

For example, through our first product, the Li ONE — at that time we only had a few thousand people — we used that product for market validation, to complete our zero-to-one phase. Though zero-to-one was still substantial, since we built a single product that did $10 billion in revenue.

This phase also includes today we're working toward a target of 1.6 million vehicles, which may exceed the combined sales of Mercedes-Benz and BMW in China, by 2025. This is our one-to-ten phase. These two phases — what a company does across several dimensions is actually different.

From my perspective, I focus on four important domains. The first is product R&D — this determines our ceiling. The second is our overall sales and service. The third is actually our overall supply and manufacturing. Internally we call these value creation, value delivery, and value realization.

These are our three business domains. Why do we split them into three? Because it's too complex. Manufacturing includes procurement, supply chain, our parts suppliers, so we divided it into three domains.

The fourth thing we especially emphasize is overall organizational capability — how to organize all these people.

In the first phase, I think for product R&D the most important thing was product definition, because you still didn't know if these things were definitely right. So we didn't blindly invest heavily from the start and self-develop everything. A lot of times we relied on combining resources and a small amount of self-developed technology to build a product like the Li ONE.

In sales and service, the most important thing was product communication, getting people to understand our product. At this point branding didn't have much effect. In supply and manufacturing, we built our own factory — we invested several billion to establish the Changzhou factory, closing the loop on vehicle quality.

On the organization side, the most important thing we did was human resources. Why human resources? Our people came from all kinds of different industries — automotive, AI. In retail we hired a lot of people who had worked at Starbucks, at IKEA. These people had all been using different languages.

Tuobuhua: And bringing successful experience.

Xiang Li: And they had never even worked together before. Now we needed everyone to work together, so human resources became critical — including how to have a common language, including later when we rolled out OKRs, how to get everyone aligned on goals. Otherwise if they just stuck to their original habits it wouldn't work. That was the first phase.

In the second phase, starting last year, platform R&D came into play. The L7, L8, L9 and vehicles to come — the entire electric drive system behind them, the entire intelligent driving system — are all consistent. Platform capability became critical.

Second, on the commercial side our sales capability became critical. We need the ability to sell that many vehicles — it's not easy, and we're direct-to-consumer.

On top of that, scaling up the entire supply chain became critical. Making one vehicle and making ten are completely different challenges, so what capabilities the team needed to build at this point mattered enormously. On the organizational front, what mattered most was process — and I mean process at scale, because you're becoming tens of thousands of people, you're fighting like an army group. Phase one was guerrilla warfare; phase two is army group warfare, and that's when process becomes crucial.

What counts as process in my view? Organizational process is the product in my eyes — it productizes our various best practices.

Tuo Buhua: That's beautifully put. Process is product.

Xiang Li: Right, process is product. It essentially productizes all the best practices across our business, so everyone sees the same thing, rather than each person running on their own experience.

To put it plainly, process is the most important product we offer our employees. If they all know how to do PEA, the quality of everything we produce becomes extremely high — it's as if everyone can easily use this vehicle, rather than having lots of unwritten rules and hidden expertise behind the scenes. That doesn't work.

Tuo Buhua: This is actually quite counterintuitive, because I've met countless founders — Dedao has many entrepreneurs of all sizes among its users — and when it comes to making products, delivering, solving real problems for end consumers, everyone is incredibly serious.

But when it comes to process, it's lip service. In principle, yes, we need process development, organizational development. But when it actually comes down to process, especially as a founder, you find that I can bend the process, or not fully follow it, or skip it, and it seems fine.

But this concept you just described — that process itself is product, that you need to approach process with the same seriousness and rigor you bring to delivering for consumers — this is actually, from my own experience, a major cognitive shift for founders.

Xiang Li: It's not necessarily that those founders are wrong. I think it's different stages. Because in phase one, guerrilla warfare is the highest efficiency. If you start with a regular army, you might end up as warlords.

So at the very beginning, in the early stage of scale, guerrilla warfare is the most effective approach. Guerrilla warfare actually requires flexibility. Process is like — guerrillas definitely run on dirt roads, on ordinary small paths, but eventually a regular army must run on highways. The requirements behind these are different. Someone has to build the highways. The highway is the product of your entire transportation system. That's our view.

I think more systematically, what is it? Don't mismatch stages. If we hadn't been guerrillas early on, we'd be dead too. Because, as we often say, many companies don't have Apple's fate but catch Apple's disease — this is the root cause of death for many companies. You need stage-appropriate approaches. Mainly because when we reached the stage of tens of thousands of people, hundreds of billions in revenue, what problems did we actually need to solve?

Originally intuition could solve everything; later, intuition and reason had to be separated.

From a rational perspective, we further divided into user-facing and employee-facing. What's user-facing? Making good products. That's the most rational thing. Even if you're Mercedes-Benz, BMW, or Audi, if your product is bad, people won't buy it — your brand is definitely not the problem. So this is our contract with users.

What do I think is the contract within the team? It's actually organizational process — it binds everyone together, and it must be explicit, not a pile of unwritten rules and accumulated experience.

When you get big, it's like building highways. Everyone knows how to drive on highways, and everyone basically knows the speed limit is 120. Although highways have many rules, the transportation efficiency of highways is definitely the highest — no developed country lacks highways. And highway accident rates are also the lowest. It's just that when accidents happen they're slightly bigger, but the accident rate is far lower than normal roads. That's the first benefit.

What's the second benefit? It gives everyone a training-and-combat mechanism. As long as you drive on one highway, you'll find you can quickly drive on all highways — you don't need to learn again how to run on mountain roads and dirt paths, how to run in swamps, how to run in deserts. So it gives everyone a training-and-combat mechanism, including the PEA we mentioned earlier.

PEA is small-scale; there are larger things like IPD behind it. Everyone can truly train and combat within it, breaking away from the traditional master-apprentice model.

Tuo Buhua: Including experience.

Xiang Li: Right, so I say this is the second point. The third is that it creates a contractual system. It's different from how we used to do OKRs — it's actually an interlocking mechanism.

We need multiple teams to collaborate. It's a contract between teams, between team and individual, and between team and company. And this contract is two-way — both sides must fulfill their mutual commitments.

So under this contract, it's quite interesting. Does the company become less efficient? Not at all. It presents something particularly rational: first petty, then noble. When setting targets and allocating resources, people argue fiercely. To put it bluntly, I can sell 300,000 vehicles, but if you can only manufacture 200,000, then I've wasted my effort at this company. So you have to manufacture 300,000 too.

If you don't have enough resources, we can allocate resources to you. So it becomes: when setting targets, the targets are very detailed, and people argue with each other, argue to the heavens. But when it actually comes time to execute, they're solid as iron. To put it plainly, from a rational perspective, we can both make good cars for users and make good processes for ourselves. I think this is the rational level.

At the emotional level, what's user-facing is brand; what's employee-facing is our culture.

Brand actually works subtly. People who like using iPhone — in whatever they do, it's simplicity, they do subtraction.

We at Li Auto create mobile homes, create happy homes. I guarantee that 99% of our users, after driving our cars, will have better relationships with their children and parents. It works subtly. This is emotional guidance.

For example, many of our users reported that the father at home didn't used to pick up the kids much. Since buying the car, he's willing to pick them up every day, and his relationship with the children keeps getting better. That's emotional.

Internally, it's culture — when everyone believes in the same things, it helps us succeed.

Many times in dark areas, when there's no road, as long as you do this, people develop mutual trust. So we say, what reason actually solves is credibility; what emotion solves through brand is trust. Users trust your brand; employees trust your culture. So they collaborate together.

These two are complementary.

If a company's culture is bad and it just goes and builds process, that process becomes something a few departments use to make things convenient for themselves while harming others — so the process becomes complex and rigid.

Tuo Buhua: It becomes bureaucracy.

Xiang Li: Right. If the company culture is good — creating user value, truly collaborating, having these capabilities in its bones — its process is simple yet rich. Because everyone can quickly see that process is just standardizing my work, my best practices, productizing them, distilling my capabilities.

It's like building various roads, laying out highways more and more comprehensively, making everyone's transportation efficiency higher and higher. I think this is what we mean by treating organization as a product.

Tuo Buhua: I think after you used the highway concept, it becomes very easy to understand. The road network gets denser and denser, but the rules actually remain simple. I don't need to learn a new set of rules just because I get on another highway. After I learn the basic rules, I can use the entire network's connectivity.

Xiang Li: Including that many small companies can't do job rotation, but when companies get big, once these roads are built, people can rotate.

Tuo Buhua: Job rotation becomes necessary.

Xiang Li: People in sales can do HR; people in HR can do finance. Why? Because there's a system in place.

And at this point, if someone only knows the business, they might only grasp one-third of the entire management world. But if they also understand HR, they grasp two-thirds. If they understand finance, they're a qualified operator — they can see the whole picture. When they look again at business problems, HR problems alone, finance problems alone, they'll find them easy to solve because their dimensionality is different.

Tuo Buhua: So if you reach this point, as a company, you've also delivered on employee growth.

Xiang Li: Right. Employees won't hit a ceiling in one position and never move up. So we can see that all the world's top companies do job rotation — it's because they've built such a training-and-combat system.

Tuo Buhua: You put that beautifully. Since I started my own business, the best advice I've received was from an investor who reminded me: "Tuo Buhua, don't tell me about your company's special characteristics. Small companies have characteristics; big companies don't. Big companies must grow to look a certain way, because they need to use resources at that scale and manage organizations at that scale — they simply must grow that way." It's exactly what you said: big companies must build road networks and establish the most efficient rules of the game.

You described actually producing two things. One is producing credibility — your trustworthiness. The other is trust — people's willingness to believe in you, and when standards aren't clear, using culture, brand, and philosophy to compensate for rules and processes not yet formed.

I think this is remarkable. And in the course, you also detailed how you distilled your culture — that culture isn't slogans, isn't shouting an adjective. That's not culture. It must be a set of behavioral rules, and these behavioral rules are about choosing what's suitable for you from among many correct options.

So people often say our culture, our values are integrity — well, is your antonym dishonesty? No one would choose that. It must be choosing from among many correct options the one that can best get things done, that best fits our team culture, and that best fits our user value creation.

Xiang Li: Or more importantly, the way to distill it: what is culture? Culture is what you truly possess that can help you continuously succeed — never make it about filling gaps.

Culture of filling gaps — all of them have big problems. For example, some company's culture is: innovation. Including many investors saying this company's culture is "supreme goodness is like water."

Tuo Buhua: Basically the relationships aren't very good.

Xiang Li: We know this company actually has a big problem with innovation, or that people at this company don't get along well.

Tuo Buhua: Basically companies that write "supreme goodness is like water" — I've concluded the boss has a bad temper, because this is definitely the boss's self-constraint, but it's usually filling gaps. People tend to care too much about the expression when distilling culture, so-called values. But actually I see that neither Autohome nor Li Auto fussed much over wording — both were quite plain.

Xiang Li: We think what's most important about these things is shared understanding, and the greatest meaning of shared understanding is: when I say this sentence, ten people's descriptions of it are the same understanding. If ten people are different, imagine how low this company's efficiency is.

Tuobuhua: This is also testable — whether your cultural values are a good summary.

Xiang Li: Everything can be tested. For example, our user positioning is families. How do you test that? Go to our stores and look at who's there looking at cars. A lot of companies say their brand targets young people, but you go to their stores and there aren't any young people — it's all people looking for young people.

If you go to a Li Auto store, you'll definitely see people coming with their kids and parents. I think the core thing is that what's in our heads, what we say, what we do, and what we ultimately achieve — all of it has to align. Only then does it work. So everyone needs to reach consensus on this.

Tuobuhua: This is actually a reminder for us too. When we're building culture, we can also use product-testing methods to test whether the culture we're building actually helps or promotes consensus.

Xiang Li: So a lot of the time, when you watch people do summary meetings, when everyone encounters difficulties and does retrospectives — are they talking about the core of our culture? Because people can't fake it. Everyone will discover whether this thing is really useful when we're at our hardest moments. I think that's real culture.

Tuobuhua: I heard you promoted a book at your company and made an outstanding contribution to its China sales — The 7 Habits of Highly Effective People. It's actually an old book, but I heard you require everyone to read it. Is that true? Why?

Xiang Li: Not everyone. We actually worked with FranklinCovey, bringing in some of China's best trainers to help us develop our own proprietary course.

Tuobuhua: So a Li Auto version of The 7 Habits of Highly Effective People.

Xiang Li: You could say that.

Tuobuhua: Why did you choose this book to do that?

Xiang Li: Because I feel our education system doesn't really teach us how to handle all kinds of complex relationships. Actually, I think the Chinese translation of The 7 Habits of Highly Effective People isn't very good, but this book is really a gold mine. Personally, I think this book emerged during an era when success literature was everywhere. But I see it as a philosophy for how to be a person and handle affairs, and it has a theoretical system behind it that we've dug deeper into.

I think the core is that it teaches us how to handle all kinds of complex relationships, because when we enter society we must have this ability. If you lack this ability, or you're missing part of your social competence — and just don't know it — that's even more frightening.

After we customized this course, we realized that our whole lives are about training ourselves to handle three types of relationships. The first is handling your relationship with yourself — how to understand yourself, manage your relationship with yourself, adjust your condition, know what you really want, and know how to effectively arrange your time.

Another very important point is that when we encounter problems, we should prioritize looking inward for causes. Really, 90% of causes are self-created, not caused by others. It's just because your ability is poor that you always think it's because of others. That's the first level.

The second level is that we lead many teams, whether project teams or physical teams. At this point we need to handle our relationship with the team.

We need to understand every person, handle their relationships, and help each person leverage their strengths and avoid their weaknesses.

Don't always think about changing people — nobody wants to change. Actually, a lot of the time personality is determined innately. When you understand each person, you can help each person leverage their strengths and avoid their weaknesses. That's the meaning of a team. The rational and the emotional are actually a multiplicative relationship, not a you're-right-I'm-wrong relationship.

So I think we often say something particularly naive: why is he different from me?

Tuobuhua: Of course he's different from you.

Xiang Li: Right. He's valuable precisely because he's different — that's where synergy comes from. So this is the second level. If we want to become qualified managers, we must understand the team, handle our relationship with the team, and help each person leverage their strengths and avoid their weaknesses. Actually this book explains it all in detail — win-win thinking, seek first to understand then to be understood, synergize.

I think if you want to manage a bigger team, like a company or a large business unit, when we talk about how you build leadership — what's hardest about leadership? Actually it's reaching consensus with everyone in all kinds of complex, very difficult environments, helping everyone reach consensus.

So at this point you need to let others understand you, and also help others understand each other, thereby building consensus. I think consensus is a key to leadership. Whether it's Warren Buffett, Jeff Bezos, or Steve Jobs — their ability to build consensus is extremely strong.

Many people have another problem: they complain, why doesn't he understand me? As I just said — why is he different from me? Why doesn't he understand me? Actually, getting others to understand you is a higher-dimensional obligation of yours. Not only do you let others understand you, you also have to help others understand each other, thereby helping everyone build consensus.

So I think this is what The 7 Habits of Highly Effective People helps us see — how to train every person to become an adult, how to train every person to become a qualified manager, and let some people with more potential have the opportunity to become leaders.

Tuobuhua: That's remarkable. I've also read this book several times, but the process you just summarized — I think you've developed this system. In today's Chinese workplace, probably everyone has bought this book, but the number of people who've really read it and applied it may not be that large. So if everyone has this book on hand, you can look at Li Auto's practice to see how to cultivate yourself and help others.

Xiang Li: You're right. What we often say is: eating lamb or beef, you need to grow it into your own flesh, turn it into internalized ability that helps you level up — not simply copy it.

Tuobuhua: Not copying — you've really developed this system. I think this generation of outstanding Chinese entrepreneurs all have this characteristic. Xing Wang at Meituan also has his "Four Great Novels." He particularly advocates The Pyramid Principle — he says it's actually a set of rational, structured thinking abilities. He also feels that in the process of solving problems from childhood on, we didn't have structured training, so everyone's efficiency is low. So he particularly advocates The Pyramid Principle.

I think this is excellent, because it's equivalent to saying, every good company is itself also a school — it's also in the process of developing people and cultivating people.

Speaking of process design, I want to chat with you about a topic. I noticed that some time ago you posted very long content on Weibo about paying tribute to Huawei. Honestly, not many entrepreneurs would do this.

Because first, you more or less still have some competition, and you also said at the time that the AITO M7's impact on you was very significant — it was a very strong competitor. But when I saw that share of yours, I was personally very moved. You stated a fact — got beaten badly by the M7 — but the larger part after that was paying tribute to Huawei.

Including that you said your executives probably bought more than ten books each about Huawei, going to learn from Huawei's process design. I especially want to discuss a question with you: In your entire entrepreneurial process, has your relationship with competitors, or broadly with these competitors, also undergone a subtle change?

Xiang Li: Actually I don't have that strong a concept of competitors. Many people see us as competing, but actually we're just training the team.

Tuobuhua: Creating pressure?

Xiang Li: Right. Because a lot of the time, for example, when my L9 is selling well, they'll get complacent. At this point I actually need something like this.

Tuobuhua: The wolf is coming.

Xiang Li: Right. You need this kind of training to ensure the team's vitality and atmosphere. But actually for longer-term development and what results we want to achieve, I think 99% is determined by ourselves. If you're not pulling your weight, nothing else matters.

Tuobuhua: Right. Actually in the final lecture of our course, we also specifically talked about making growth your most important task, your most important thing. Everyone wants to learn; as an organization, a company needs to learn. How do you push this team to always stay in a learning state?

Xiang Li: I think two aspects. One is that all our processes, like AI, are a learning and training mechanism. Your outward observation is your learning; your planning and your final execution and operations are actually your validation after learning. Finally your retrospectives and summaries are equivalent to precipitating capabilities onto every person and process, because without retrospectives there's no growth — it's just doing whatever and leaving it at that.

All the processes we build, all the methodologies we bring in from around the world — we learn Toyota's TBP, we invented PEA ourselves, we ask consulting firms to help us design talent decoding — you'll find they're all a learning mechanism. So everyone running on this path every day is already in a learning mechanism.

Second, we have another particularly big advantage: we basically don't punish mistakes.

Tuobuhua: You don't punish mistakes?

Xiang Li: Right. Our biggest punishment for mistakes is that you have to level up and solve this problem.

Tuobuhua: You're the person responsible.

Xiang Li: Right. Although you didn't do this well, next you need to solve this through stronger capability upgrades, and often they can solve it, because they have more sense of responsibility. Not punishing mistakes actually makes people willing to learn, willing to try new things.

A lot of the time once a company's rules become heavy punishment for mistakes, everyone will feel it's better to do less — do less and you don't need to learn. As long as you keep moving forward, you need to learn; not learning, you fundamentally won't know how to move forward.

Tuobuhua: Right, this is very remarkable. Actually it's embedding learning into the mechanism, embedding it into the process, so it's not calling on everyone to go out and learn — rather, your process itself contains a learning input system.

Xiang Li: If I could add one more thing, our incentive system is also like this. In our final performance evaluation, growth is very important. We divide it into two parts. The first part looks at business growth, looks at process growth, and also looks at his own team's growth.

If we return to the individual, we look at professional capability growth, look at management capability growth, and also look at leadership capability growth. The culture itself is also built around this, and the incentive system is also built around this.

Tuobuhua: For us, continuous learning — including today coming to this course to learn from Xiang Li and Li Auto, to see how they build products and how they build the organization as a product — this is all part of our own input process. But more important is the reminder he just gave: Have you put your learning and innovation mechanism into your real processes?

Xiang Li: It needs to be a closed loop.

Tuobuhua: It needs to be a closed loop — can't be self-indulgent. For example, today I listened to Xiang Li and gained so much, tomorrow I'll do whatever I was going to do anyway. Or even that my learning has no way to be validated in my behavior — then the effectiveness of such learning is probably very, very low.

I'm personally especially moved and excited that Xiang Li could come to Dedao to do this course. But just now I saw in the comments that many people are interested in this question: today you're doing this course at Dedao — what is your motivation or your goal?

Xiang Li: I think two aspects. First, our team already validated this when we were doing PEA.

One is our external gain, one is our internal gain,

I think the external gain is that Dedao's user base is very close to the user base we want. And this user base has another kind of value — it's not about how absolutely large this user base is, but that its influence is very significant. They have the ability to influence an even larger circle around them.

Tuo Buhua: Exactly right, and they're all like me, especially eager to share.

Xiang Li: Yes, so if we can learn and explore many issues together with such a group, I think compared to other automakers' marketing, we're conducting a dimensionality reduction attack. Because the human dimension is different.

Second, in their communication with the Dedao team, Dedao can also help us close the loop on our accumulated experience over the years — it's equivalent to a training exercise. Because doing something, articulating it, and being able to withstand scrutiny from more people are three different things. I think Dedao is very capable in this regard; it's essentially helped us productize and close the loop on some of our past shareable experiences.

Tuo Buhua: Not a single sentence without the word "product."

Xiang Li: Right, I think this will also make us more effective at articulating these concepts internally. Because when I speak at the company, it doesn't require validation — I'm the CEO.

Tuo Buhua: You're a "tyrant."

Xiang Li: Whatever I say, people will probably just listen. But I think we need external perspectives and validation from professional external teams, which can also make many of the things I communicate internally more effective — it's also a form of growth.

How do you avoid letting your individual experience blind you to the real situations of a broader user base?

Tuo Buhua: The earlier questions were all quite macro and big-picture. Now we have some very specific questions. The first one is something I'm personally a bit puzzled by: you specifically emphasized that as a product manager, as the founder of Li Auto, you yourself are a typical user of automotive products, especially family vehicles. But we also know that when the boss themselves is the user, they often fall into that trap of "I don't care what you think, I care what I think" — essentially over-relying on their own experience to infer what users want. In this process, how do you avoid this phenomenon?

Xiang Li: It goes back to what I just said. First, you need a good team — a group of people who dare to speak truth and persist in speaking truth. Second, you need a validation system. Any feature I push also goes through detailed validation by the validation system.

Tuo Buhua: It seems like you have some kind of co-creation session tool with users?

Xiang Li: I don't participate in that part.

Tuo Buhua: You don't even attend the co-creation sessions?

Xiang Li: I don't participate. I think what's more important is the subsequent product validation overall, and we spend more money than ordinary companies to build various scenarios for validation. As I said, for long-cycle products, emotional needs and rational validation — neither can be missing. And in the team, there's a group of people who dare to speak truth and stick to their views; we clash with each other. At first they might be cautious, but later there's no pressure anymore. So basically, in many respects, I lose more often, and they win more often.

Tuo Buhua: This is especially enlightening for me, because Dedao is smaller in scale and we're currently going through this process. For example, in the early days, because Mr. Luo himself is a bookish person with very good taste in knowledge, often after a course product was approved, they'd bring it to Mr. Luo to take a look, and his sales estimates would be very accurate. As time has passed and our users have become more complex, we sometimes joke that when Mr. Luo says "I especially want to learn this course," we'll say that might actually be a negative indicator. Because he's just different — a founder's taste and preferences don't completely match users' real needs. So we're also currently figuring out how to build this validation system.

Xiang Li: I think it comes back to the people angle. You need three capabilities here. First, you care about user value and product more than you care about yourself — this is an especially important capability.

Tuo Buhua: Make yourself smaller.

Xiang Li: Make yourself very small — not that important. At first people didn't believe it, but slowly they discovered that when we hold review meetings, I ask a few questions, and then everyone ignores my existence. I think this is a very good atmosphere. Because you can't just pay lip service; you have to genuinely act that way — that you're not that important. What's most important is that everyone produces the result, creates the best value. Once you have this capability, you can then cultivate two other capabilities. One is the ability to help everyone reach consensus — getting everyone to reach consensus, extracting what's in everyone's minds, letting everyone see a more complete world. You can't have each person keeping their stuff hidden in their own head.

Consensus is produced through co-creation. We say even when people argue, that's co-creation. Everyone pulls out what's in their head and helps everyone see a more complete world. At that point, making judgments becomes very easy. Otherwise, everyone is standing on their own little patch of ground looking at things, and it's very difficult.

Once you have this capability, it produces a particularly important ability: the ability to reach consensus. And what is reaching consensus? Honestly, it's essentially creating a super-brain — a brain that's greater than any single individual. I think this is a particularly important capability.

Then immediately a third capability emerges, an especially important one: seeking truth from facts, and correcting mistakes when you make them. Because once you have this capability, you discover that seeking truth from facts and correcting mistakes is actually extremely efficient for team productivity. Because no one doesn't make mistakes. At this point, you unlock more positive value and energy. But the root of all this starts from the first point: that you yourself are not that important.

Tuo Buhua: You're not that important, the founder's not that important, the company's official stance isn't that important. I previously came across something called the McKinsey & Company formula. It says that to get something done, the essence is this formula: reliability × likability × credibility — it's multiplication. You first need to be reliable, then credible, then likable. At the brand level, most importantly, you have to divide by something — divide by "your ego." The bigger your ego, the smaller your capacity to bear; the smaller your ego, the more effective all the efforts above become. I think this is actually consistent with what you're saying.

Xiang Li: What's important about me? Of course I'm important, but what am I important for? What's important is getting everyone to truly focus on valuable goals. What is my importance? It's ensuring that everyone spends enough time reaching consensus, opening up their brains, and ensuring that everyone is seeking truth from facts and not deceiving themselves. At this point, what's most important is that I'm both unimportant and important.

Tuo Buhua: Yes, so I think for all of us, all founders probably have to pass this test: without your own preferences and obsessions, you couldn't have gotten started. But if you become too trapped in your own preferences and obsessions, without building a co-creation, consensus, and error-correction mechanism, then it's not sustainable.

---

With multiple product lines, all targeting family users, how do you differentiate?

Tuo Buhua: The next question I'm a little worried about on your behalf, because I think for a young company, this is a huge challenge. In the past, you made hit products — the Li ONE was great, the L9 was great — but now you've become a multi-product-line company. That's actually a capability where traditional companies are especially strong, managing many product lines, even many brands. Now that you're facing this challenge too, how are you dealing with it?

Xiang Li: I think there are two levels to this. First, when a company succeeds at one thing, after you've gone from zero to one, what can you actually do from one to ten? I think every failed company we've seen died because from one to ten, they did something completely different. What all top-tier companies do is: from one to ten, they keep strengthening what they validated from zero to one. At that point your chance of making mistakes is lower. Why? Because you don't need to trial-and-error anymore. The core is how can I create value for users more efficiently, how do I improve your team efficiency, your overall efficiency — I think this is especially important. We've indeed seen many companies that did well from zero to one, then immediately died from one to ten.

Tuo Buhua: They did something else.

Xiang Li: They did something else, and started trial-and-error all over again. They worried their original market was too small, but actually it's different from what people imagine — if you really have the ability, you can create something. That's the first point.

Second, as we just emphasized, is exceptionally strong iteration capability. Not all of our products succeed. Many of our products, if launched the way traditional companies do, probably would have failed. We're able to recognize mistakes and correct them.

For example, basically six to nine months before our products launch, we do another round of real user validation. At that point we see many problems. Often, many companies just don't fix these problems because there isn't enough time. So we have this exceptional efficiency where everyone has strong consensus — all the teams, design, engineering, product, and the manufacturing teams behind them — we'll spend enormous costs to make changes. Because maybe I spent 50 million, even 100 million, 200 million on the fix, but ultimately it brings me revenue in the hundreds of billions, so it's incredibly worth it.

But if I don't fix this thing, maybe my sales drop from 15,000 to under 1,000 — that loss is immeasurable. Many times companies think, if I change it, doesn't that prove I was wrong before? I don't actually think that way.

Tuo Buhua: Isn't proving you were wrong a kind of gain?

Xiang Li: Yes, so in this regard our iteration capability and reaction speed are extremely fast. This capability is written into our blood and our DNA.

Tuo Buhua: That's remarkable. Being honest with yourself is truly remarkable. Honestly, it's very easy for people to fool themselves. I remember you talked about a remarkable case in your course — the matter was small, but I believe the cost was high for you too — the dual armrest issue.

You had already announced it, said the Li ONE had dual armrests, but during validation discovered the outer armrest would snag clothes, would poke legs. So already announced, even promoted to users as a selling point, but at delivery removed one armrest. This kind of error correction, I think, isn't easy.

Xiang Li: This isn't even the most extreme. What's most extreme is: when we were making the L9, we were basically the first Chinese company that wanted to do full-size SUVs at scale. In the past, the two products that did full-size SUVs best were the Mercedes-Benz GLS and the BMW X7. Their door designs were all half-wrap doors — when you get out, there's a sill, it somewhat rubs against your legs. So after studying them, we thought half-wrap doors were probably a universal approach, because they'd been validated.

But after we launched our car, we discovered the half-wrap door experience was terrible. It made getting in and out extremely painful for kids and elderly people. And we're somewhat different from fuel cars — because I have batteries underneath, my sill width is wider, making getting in and out even harder.

At this point it was already February 2022. The car was launching in July, delivering in September. At this point no car in the world would still be making changes. It was still during Spring Festival. But we made the most important judgment: change the doors from half-wrap to full-wrap. Normally this kind of engineering takes 12-18 months. What does that mean? It means my doors need to be redesigned, my entire side roof needs to be redesigned, and my production line needs to be redesigned.

Tuo Buhua: Yes, I can imagine.

Xiang Li: The total investment was about 60 million. We still had to change it, otherwise this car would fail when launched. Because as a family car, if I can't even guarantee something as basic as elderly people and kids getting in the car, how can I sell it? So it's not that this product was successful from the start — actually a lot of work was done behind the scenes.

Tuo Buhua: Truly remarkable. Spending 60 million to fix one error — that's genuinely one of the most expensive errors I've ever heard of. Of course, if you don't fix this error, if you're not honest, that loss is incalculable.

Xiang Li: Right, I think it's a loss in the hundreds of billions.

Tuo Buhua: Even whether this company still exists would be hard to say.

Xiang Li: Yes.

Tuo Buhua: It's still that same point — being honest with yourself remains a lifelong practice, especially for founders. Because fooling yourself is the easiest thing. I think founders need to be especially vigilant, because after reaching a certain stage, you'll discover the whole world is fooling you. Because as long as you don't correct errors, actually everyone has it easy — but the accumulated result is often something we can't bear.

So I think you need to build a mechanism for how to prevent an error from flowing to the next process, and for it to be resolvable when it should be resolved.

---

How did Li Auto "pinch" out its profits?

Tuo Buhua: I don't know if you'll mind this next question. At least in Beijing, in big-tech circles, having a nickname is basically a badge of honor. For example, Tencent's nickname is "Goose Factory," Baidu's is "Wolf Factory." Do you know your company's nickname?

Xiang Li: No.

Tuo Buhua: Then let me tell everyone — in big-tech circles, Li Auto has a nickname: "Pinch Factory," as in stingy. The consensus is that Li Auto is especially stingy, extremely tight on costs, very tight-fisted. Including that you yourself wrote a long Weibo post about why there isn't that much PR budget, marketing spend — honestly, very few people directly confront this issue.

Of course, this "Pinch Factory" reputation is also, from another angle, an honor. It shows your cost control is very good.

But honestly, because you yourself built PCPOP, built Autohome — theoretically you didn't have this experience, because frankly, pure internet platform companies or content companies still tend to be relatively loose with money, not that strong on cost consciousness. Why after founding Li Auto did you become "Pinch Factory"?

Xiang Li: I think it's different from what you said. Autohome was a company with extremely high profit margins and efficiency.

Let me give you a number. Autohome's entire forum accounted for roughly 95% of automotive UGC content. But our headcount was the smallest — we only had about a dozen people managing this forum. Our competitors had only 1-2% share, but needed to support two to three hundred people to manage it.

Tuo Buhua: Efficiency.

Xiang Li: Otherwise you can't prove you're a tech company, I think that's one aspect. Let me give another example. At Autohome, there were over 30,000 dealers nationwide. These 30,000+ dealers needed to continuously pay us. Our finance staff was 0.5 people — he mainly did other things, maintaining dealer finance part-time, also handling invoices. But our peers basically had finance teams of 50+ people.

Tuo Buhua: Then I'm very curious how you achieved this. This is an order-of-magnitude difference — it's not achieved by pinching pennies.

Xiang Li: Right, because stacking people is an especially easy thing. So often when you say don't stack so many people, don't spend so much money, actually people can think of many better methods. Otherwise what is all this technology for today?

Technology isn't decoration — technology is productive force. For example, can these invoices be automated, and even when picking up and delivering these invoices, can SF Express's system connect with ours — there's a lot that can be done behind this. When you have this kind of requirement, people go and do innovation in this area.

Another level is: in 2008 we encountered an economic crisis. At the time Autohome was growing extremely fast, roughly 7-8x the previous year's revenue, but advertising had 6-9 month payment cycles. So when we hit the economic crisis, our cash flow broke — that was extremely painful.

You can see that from when Li Auto officially started deliveries in 2020 to today, our cash flow has always been positive. Tesla hasn't even achieved this. Behind this is also your pursuit of efficiency.

So I tell the team, the auto industry is a very tough industry — we need to climb up from the 18th level of hell, not start from ground level or even from the tenth floor. So when you reach ground level, your competitiveness will be completely different from how others trained.

Tuo Buhua: Essentially, I think for all companies today, pinching costs and pinching profits is definitely extremely important — otherwise you can't survive. The entire market environment is also forcing everyone to do this. I especially recommend that if you're interested in this issue, you can listen to Xiang Li's 15th lecture: profits aren't pinched out by being stingy, but are indeed pinched out by pinching processes. He specifically mentions using an "atomic-level" way to understand costs, understand profits, and truly put resources where they most need to be used.

---

Why emphasize retrospectives?

What is the most important capability?

Tuo Buhua: I have two final questions about your personal working methods — both somewhat unusual. The first is about what you just mentioned, your strong emphasis on retrospectives. I understand this is also part of Li Auto's culture, and apparently quite different from how most companies approach retrospectives. Could you elaborate on why you value them so much, and what's the most important capability within that process?

Xiang Li: If I had to put it this way, I want this company's growth to be like climbing stairs, not like high jumping. I need to make sure that when I step onto the next stair, I'm standing at the height of the previous one. What's the next stair? And what's the height of the previous one? That comes from retrospectives.

The things we do well, when we institutionalize them — that determines the height of each stair. Once I have my foot on that stair, it's mine. I won't fall back down. Then there are the problems we haven't solved at this stage. How do we prioritize solving them in the next stage? Finding the solution is how I step up to the next stair. Where exactly is my next stair? That's how we think.

One crucial thing about our retrospectives is that we need to capture the capabilities that led to our successes — and for the things we didn't achieve, I need to figure out what to improve. Was it my insight? My planning? Or did something go wrong in execution? We can always analyze that.

Tuo Buhua: So your retrospective meetings are very substantive?

Xiang Li: They're a necessary part of the process. It's just like training AI.

Tuo Buhua: What role do you personally play in these retrospectives? Facilitator? Initiator?

Xiang Li: Most importantly, I need to make sure it's done seriously, not just going through the motions.

Tuo Buhua: So you're accountable for the retrospective process itself.

Xiang Li: Right. We have to retrospect on what we did well, so you understand why it went well — not just luck. We also retrospect on what didn't go well. There's no difference between the two. If you solve what went poorly, that's still a gain.

Tuo Buhua: Yes.

Xiang Li: Retrospectives are a habit I've had since I was young. I discovered their huge payoff in high school.

Tuo Buhua: Was there some particular moment? Because very few people become aware of this so early.

Xiang Li: I think it's fundamentally basic. I've always believed that the "thesis–support–conclusion" structure we learned for essays as kids is excellent. The first "thesis" is stating the problem you need to solve. The "support" is how you go about solving it. The "conclusion" is where you retrospect on whether you actually solved it. I found it incredibly effective — a very simple structure for getting things done.

Often the middle part is relatively easy to handle. But whether the problem you identified upfront is the real problem — that's a challenge. And whether you verified afterward that your process actually solved it — that's a crucial challenge.

Tuo Buhua: Retrospectives are honestly important, and simply doing them isn't easy. But even when people do them, many stay at the surface level. Without that thesis–support–conclusion structure, there's no closed verification loop. Because in a retrospective, even with a very serious attitude, I might still fool myself and others.

How do you verify whether you've actually done it or not? Because after you complete something, there might be a hundred metrics you could present. I've seen many retrospectives where people just cherry-pick the metrics most favorable to themselves.

Xiang Li: You mentioned Christensen's book Competing Against Luck. I think retrospectives turn luck into capability.

Tuo Buhua: That's well put — turning luck into capability. But many people's problem is they can't even tell when it's luck and when it's capability. That's the real pity.

Xiang Li: Seizing the right timing is itself a capability, not luck.

Tuo Buhua: Exactly. So how to continuously use retrospectives to analyze your capabilities, summarize them, and institutionalize them so you can step up to the next level — I think this is crucial whether you're a founder or an ordinary young professional.

Our whole life is a process of climbing one stair after another. I find this particularly characteristic of Li Auto, and it's been enormously enlightening for me personally.

---

How to Efficiently

Integrate New Employees into a Collaborative System?

Tuo Buhua: There's one more thing — probably no one else at a 20,000-plus-person company does this. You mentioned that for all new employee training, you give a talk every single week.

Xiang Li: I attend every new employee training session.

Tuo Buhua: That's incredibly time-consuming. Can't you just record it for people to watch?

Xiang Li: I think this is the most important thing. When a new person joins our team, if they form key consensus with us from the very beginning, their efficiency in understanding and integrating into the company becomes extremely high.

If all of the company's principles — including our value proposition to users and our value proposition to the organization — are things they have to speculate about, to guess at, that's a terrible experience for them. Put another way, I believe I have two user groups.

Tuo Buhua: Two users.

Xiang Li: One is consumers. One is employees.

Tuo Buhua: That's the discipline of being a founder.

Xiang Li: Right. I need to reach consensus with employees, to help them better understand this company rather than coming in and guessing, reading between the lines — I think that's actually the first priority.

Tuo Buhua: Truly remarkable. After taking this course, I think the biggest lesson Li Xiang has given me is this point: if you want to get things done, you need to recognize that you have two user groups — whether as a manager or as an entrepreneur.

One is the consumer you serve — how to create higher-value products for them. The other is your employees. And he's also found what satisfies employees' needs: making everyone more efficient through process, so work involves less internal friction. At the same time, he's identified the most effective nodes in that process — new employee training being one very effective node.

Xiang Li: Extremely important.

Tuo Buhua: Time spent here probably outweighs spending more time six months after they've joined.

Xiang Li: Yes. And this does require trade-offs, because I don't have three heads and six arms — I only have so much energy. Compared to many CEOs, they might choose to attend more external meetings, which also consumes a lot of time.

My choice is to focus on key new employee and leadership training. The energy required is the same. If I choose this, I believe it's more important, so I give up that other thing. So basically over the past three years, you've rarely seen me at external public forums — I've declined them all.

Tuo Buhua: Your WeChat signature even says so — no public forums.

Xiang Li: Right. Rejecting the first one is hard. When you've rejected the tenth, everyone knows that's just how you are.

Tuo Buhua: Yes, once it becomes your signature, it's even less painful — people see the signature and drop it, no need to ask. I should give you positive feedback too: after taking your course, what you do with new employee training was very inspiring for me. I also give talks at our new employee training, but through your course, I iterated my presentation. My old deck was mission, values, explaining our culture. Now my latest version is all the mistakes we've made in our history — in the new employees' first week, I make sure to tell them about the famous mistakes we've made from founding to today.

Xiang Li: It's become a retrospective.

Tuo Buhua: Right. I just gave the first one this month. The new employees loved it — they were delighted. First, they felt no burden, thinking they'd already heard your most embarrassing side right after joining.

Second, what I really wanted to tell them is that the vision and values we emphasize today are essentially learned from mistakes. There's nothing embarrassing about these mistakes, nothing that can't be shared with others. On the contrary, it's something the founder should talk about openly and extensively. This way, for our employees, their process of understanding these things becomes more authentic, and I believe they also understand them more thoroughly.