MaHui Entrepreneur | Narwal CEO Zhang Junbin: Understanding Human Nature, Leaving Naivety Behind
What's more important right now is getting the business right and building a solid product.
- This article is republished with permission from Yingke (硬氪), written by Huang Nan, edited by Yuan Silai.
At 36, Narwal CEO Junbin Zhang realized he could no longer afford to be merely a gentle person.
He had made some unpopular decisions, but more often, Zhang struggled to deliver harsh words — so much so that some colleagues privately called him indecisive. "Inside the company, they say if Brother Bin's feedback on your work is 'not bad,' that means it's actually terrible," he jokes.
For years, this temperament hadn't exposed any critical weaknesses. Zhang spent most of his time immersed in products, maintaining a comfortable distance from tedious and complex operational matters. He loved what he did, and had carved out a corner of pure focus for himself.
But a product manager is not a CEO. A successful leader can delegate, yet never let the company slip from their grasp — and must be able to act decisively when the moment demands. "I loved doing product too much. I was too gentle, only doing what I enjoyed. I wasn't a qualified CEO," Zhang says.
During those years when his head was down in product development, problems began accumulating inside Narwal. If you were there, you'd see departmental infighting, constant turnover in points of contact, veteran employees leaving one after another. These issues seemed scattered, but ultimately pointed to a single cause: no one was capable of binding the entire company together from top to bottom.
Eventually, the root cause landed squarely before Zhang. Shocked and angry, he also came to a clear realization: human nature doesn't always follow good intentions. He began reflecting on his own absence. "When you say yes to everything, you're hurting everyone — and hurting the company."
Starting this year, he stepped out of his comfort zone, plunged into the front lines, and began using his hands as CEO to restructure marketing, R&D, and every other department. In a sense, this is the most unfamiliar and unavoidable lesson of his ten-year entrepreneurial journey: understanding people, judging people, and when necessary, making decisions about people.
Now, Zhang feels less anxious — because the business is showing clear signs of improvement. Overseas markets, for instance, saw revenue grow nearly 50% year-over-year in the first four months of 2025.
Yet the greatest shock to Zhang wasn't the business itself. After ten years of entrepreneurship, he is truly confronting the abyss of human nature for the first time. For example, when people with fundamentally good intentions do terrible things for all sorts of legitimate reasons — how do you decide? And how do you face them?
In real life, there is no absolute hatred or absolute understanding; people wander continuously in gray zones. When we asked: Has your understanding of human nature changed? He fell silent for a long time. "I now understand a saying," he finally replied. "Of all virtues, filial piety comes first — judge it by the heart, not by the deed. Of all evils, lust comes first — judge it by the deed, not by the heart."
Zhang's fundamental nature has never been cold — many describe him as decent. Yet in brutal commercial competition, a CEO need not become an "asshole," but must become a "tough person." "Mercy does not command troops" is a lesson Zhang must adapt to and learn.
We spoke with Zhang for three hours about his journey from product founder to mature entrepreneur. He is wading through this dark river alone — and still loving life after seeing its true face.
Below is the full interview between Yingke and Zhang Junbin, edited for clarity:
01
Source of Fear: What I Fear Most Isn't a Low Score, but Having No Path Forward
Yingke: You've kept a low public profile recently — Narwal didn't attend AWE this year. What specifically has been happening internally?
Zhang Junbin: Busy with marketing reform. Skipping AWE was the result of our internal review.
You need to be clear about your purpose for any trade show. Otherwise it's just burning money, and exhibitions drain enormous team energy — dozens of people spending months on booth design and location prep, for limited value.
I think there are two valid reasons to attend a show: first, when your brand is entering an entirely new market and needs concentrated exposure to media and channel resources; second, when you've released a generation-defining product or technology and need to efficiently concentrate traffic to amplify buzz.
For Narwal at this stage, we want to put resources and energy into deep localized operations. It's a question of value prioritization.
Yingke: What changed in your work after you started managing marketing?
Zhang Junbin: I used to weigh product extremely heavily — the vast majority of my energy went to product and R&D. When I first took over marketing last September, almost all my time went there.
Recently I've adjusted again — roughly 30% on marketing, 30% on overall group operations, 30% on product. The reason is that once marketing enters a new organizational phase, aligning front and back ends requires an operational lever at the group level.
Narwal's 2025 product launch roadshow, Provence, France (Source: Company)
Yingke: What was the moment that made you realize marketing mattered? Because previous conversion results fell short of expectations?
Zhang Junbin: I genuinely didn't understand marketing before. In my mind, it was just three crude moves: make it cheaper, pour more money into ads, or the product itself is flawed. I seriously wondered whether the product was the problem. Only later did I gradually realize — the issue was marketing. I can't say I understand it well now, but at least I know not to apply those three crude moves anymore.
You could sense something was off from small things. For example, calculating whether a marketing action was profitable — the team didn't even have unified internal metrics. That itself was a problem. One approach was "sell-through口径": as long as goods were pushed to distributors, it counted as sold, as completed sales, so账面利润 naturally looked good. Another was "sell-out口径": only when users actually paid and received the product did it count as real成交. Under sell-through, the numbers looked glossy, but inventory might still be stuck in channels with low turnover efficiency. So from a genuine operational efficiency standpoint, you can't use sell-through to measure monthly profit.
Then there's expense amortization rules. For a single large marketing expense, do you book it fully in the month incurred, amortize quarterly, or amortize across the full year? Without unified standards, the result is one month suddenly bearing expenses from multiple prior months, and that month's profit collapses — yet the corresponding marketing activity didn't even happen in that month.
I hadn't managed marketing before. But looking at these things from first principles, many problems weren't hard to spot.
I used to think product and marketing followed additive logic: "product + marketing." Make the product good enough, and results wouldn't be too bad.
Now I understand they follow multiplicative logic: "product × marketing." However good the product is, if the marketing短板 isn't filled, great products miss their window.
Yingke: Why couldn't you spot and solve this before?
Zhang Junbin: I was too absorbed in doing product. This fundamentally stemmed from my own immaturity — I wasn't a qualified CEO, only doing what I was good at. Only when too many things felt off did I have to spend more energy learning and changing.
The national subsidy program brought incremental growth and profit in the past two years, and growth can mask many problems. After subsidies receded, internal organizational problems surfaced more intensely, triggering our determination to reform.
Yingke: Was there no solution besides managing it personally? A CEO's mature state should be running the company through systems, doing top-level control.
Zhang Junbin: It depends how fast you want the reform to go. Founder Mode is fastest. And after some core personnel changes, if the founder doesn't step to the front, team morale destabilizes. This isn't just business reform — it's organizational reform. You need to be on the front lines transmitting positive signals, keeping everyone moving forward.
Yingke: How fast do you want it?
Zhang Junbin: Domestic reform took about half a year to basically resume year-over-year growth. I used not to understand: offline has much lower return rates than online, so why wasn't it profitable? After taking over, it's now turned profitable. In the first four months this year, compared to the same subsidy period in 2025, the company achieved revenue and gross margin growth while reducing marketing spend. That's the result of reform and adjustment.
Yingke: Fear-driven and passion-driven are quite different. When you first took over marketing, what was your fear mainly about?
Zhang Junbin: The fear came from the unknown — from not understanding marketing.
But the company pushes you to that position, so you must leave your comfort zone and reform as fast as possible.
At the time, I didn't know if we could turn it around. I wasn't afraid of the business being at 40 or 50 points — I was afraid of walking in to find everything already at 85 points, efficiency already极致, yet still no improvement. That would mean no path forward.
The reality was there was huge room for improvement, so the fear diminished greatly. Going from 40-50 to 60 is relatively easy. Facts proved this: by March-April this year, Narwal's front-end profit improved, and I became less anxious.
02
As CEO, Don't Try to Make Everyone Happy
Yingke: When did your personal cognitive growth accelerate most?
Zhang Junbin: In 2024, Narwal launched divisional reforms, cost optimization, efficiency improvements — a series of moves that faced considerable internal resistance. This required pushing things forward skillfully while guiding people to follow the strategic direction without emotional baggage.
Many decisions, stated directly, the team might not understand or empathize with. My biggest growth then was realizing hard pushing doesn't work. If you can't get everyone aligned, either people go through the motions, or worse, they try to prove you wrong.
Once organizational reform loses上下同欲, the results are often brutal. Execution deviates, final results suffer, and leadership mistakenly thinks the reform direction was wrong. It wasn't.
Solving for "shared desire" is key. All key people must reach consensus and agreement, otherwise it probably won't succeed.
Narwal's Zhang Junbin (Source: Company)
Yingke: How do you get key people to agree — through force, punishment, or incentives?
Zhang Junbin: There are many methods, but the core is making the底层收益 to the group extremely clear. You can't vaguely say "it's good for Narwal" — be specific, and let people see tangible benefits.
For example, if return rates drop by one percentage point, net profit margin increases by 0.5 percentage points — put the numbers on the table and they become persuasive. If you just say "our return rate is too high, we need to cut it," the other side might push back: "Our positive review rate is already number one in the country, what more do you want?" At that point, you're stuck.
Yingke: So you still need trustworthy people in key positions to make a CEO's job easier. What kind of person is worth trusting?
Zhang Junbin: I've thought about this, and discussed it with several top operators in China. Recently I've been focused on marketing talent. I believe an excellent marketing operator needs five capabilities.
First, they understand finance, can read financial statements, and are sensitive to data definitions — they know which metrics reflect reality. Second, they deeply understand their own product, actively using it, studying it, mastering it. Third, they can achieve big results with small budgets, not just throwing money around, but working smart with unique approaches. Fourth, they have channel resources and connections that can genuinely open up markets. Fifth, they can lead teams and manage people, organizing others to work together.
Yingke: How do you find such people and build trust with them?
Zhang Junbin: Above these five capabilities, the most important thing is motivation — why is this person joining the company?
Marketing professionals' motivations typically fall into three categories. The first is resume-padding: spending whatever it takes to run big projects, claiming credit while pushing risk onto the company. The second is resource-hoarding: using the platform as a stepping stone. The third is building something: truly growing with the company, obsessing over efficiency and costs as if they were their own.
You have to learn to identify these. Wrong motivation, however strong the capability, is a hidden danger. When talking with them, if someone keeps bringing up how they spent hundreds of millions in the past, how massive their campaigns were, but can't explain why they spent it or what real conversion and user value it created — I'm very cautious.
Yingke: But people change. Are there institutional solutions?
Zhang Junbin: I now basically attend frontline marketing operating meetings every week. I used to review things once every month or two; now it's weekly, whether domestic or overseas. I focus on four things:投放 efficiency, scale, revenue, and profit. This is essentially institutional control — not just glancing at things, but forming an institutionalized monitoring rhythm through high-frequency operating reviews.
But institutions treat symptoms, not root causes. If someone's motivation changes, and institutional monitoring detects anomalies, you have to make the change. This is no longer a capability issue — it's a character issue.
I've seen some absolutely top-tier companies with very mature processes and systems still stumble. Institutions can filter out some opportunists, but institutions have blind spots, especially in core positions.
Institutions are the floor; judgment of people is the ceiling. These things don't show up in KPIs, and systems can't run them out.
Yingke: Can you accept mistakes happening?
Zhang Junbin: Yes. The reason is that I clearly understand that a founder's growth, especially the comprehensiveness of end-to-end knowledge, takes time to grow from inside out. From R&D, product, projects, supply chain, even logistics services, to marketing — understanding the underlying logic of all these businesses takes time to develop.
Yingke: Ultimately this series of reforms comes back to values. How do you understand values now? Compared to ten years ago, how has your understanding changed?
Zhang Junbin: I've gone through several stages. The first stage was learning from and copying big companies, with very little results. The second stage, I began to understand myself, knowing what I wanted, but still unable to distill it precisely. The third stage, gradually moving from fuzzy to clear, realizing some values are essential and can truly drive company success.
Applied to specific management work, such as talent selection: in the past I weighted individual capability heavily, focusing on skills and experience. Now I feel values matter more than capability.
Self-drive and commitment matter more than capability; innovation matters more than experience. Whether someone wants to do things well, whether their self-drive is strong enough — in a sense this matters more than capability itself. Someone fully committed with 80-point capability may deliver better results than a 90-point capability person who's not all-in.
Second, speed of iterative growth matters more than experience. The core manifestation is: is this someone who looks inward for solutions? When encountering problems, do they first reflect "what did I not do well," rather than blindly blaming external factors? Only people who look inward can achieve high-speed iteration. Those lacking this trait easily become sources of internal team friction, constantly passing the buck.
You could understand this as big ego versus small ego. I now lean toward hiring the latter. Capabilities can grow, but willingness to fully commit, with self-drive and commitment to get things done — this is what I value most now.
Yingke: A very practical question: the company is in its tenth year, and this inevitably involves talent incentives and exits.
Zhang Junbin: First, Narwal's IPO process will continue to advance. It's just that last year the company encountered some issues in marketing that required reform — this doesn't mean we've abandoned going public. In this process, some people have indeed left, which is unavoidable.
Previously I talked with the founder of a well-known domestic car company. I asked him: "Your company has lost many people, and at the hardest times there were rumors you were going bankrupt — how did you retain people?" He said you can't retain them; it's the new people who joined later who carried things forward together.
Often, what companies need to do is persist in doing the right things, continuously taking responsibility for those who stay, while letting new joiners quickly demonstrate value. We won't actively push people out, but some people have been at it so long their "health bar" is depleted — they're genuinely hard to retain.
So for me, energy shouldn't be consumed on emotions like sadness, tears, and complaints from people leaving. Rather, I should think about how to quickly push Narwal to a new height.
Narwal is gradually forming its own unique culture. What makes me feel grounded about organizational reform is that new talent joining has good capabilities and energy. How to give this new team good soil to go all-out and truly do well what wasn't done well before — this is what I need to adjust my mindset to do.
Yingke: Were there times when your mindset was particularly bad? How long did it take to adjust?
Zhang Junbin: Too many. The past six months have been one such period.
This year I opened several "blind boxes" and made several key decisions. For me, organizational changes, cost optimization, efficiency improvements — all these actions — what the results would be was uncertain at first. Before results fully emerged, anxiety was inevitable. But as these adjustments gradually landed and effects slowly appeared, the anxiety eased.
What affected my mindset most was: after making so many major decisions, did the company get better? If after doing so much, results were worse than before, I'd definitely be anxious. But from results so far, things are going pretty well.
Yingke: New people need time to adapt to the organization, and different approaches easily create internal understanding gaps. How do you align this cognitive misalignment so everyone pulls in the same direction?
Zhang Junbin: The essence is granular breakdown from strategy to execution, and also overcoming some organizational inertia.
The execution layer will feel: you originally wanted to invest 2 million, now it's only 1 million, but you still want 1.5 million in results? This process will definitely create misunderstanding.
My feeling is that cognitive misalignment can't be expected to align in one shot. Its essence is caused by different information levels and performance pressures.
The alignment methods mentioned earlier include several aspects. First, goal alignment — thoroughly explaining the operating logic and achievable path behind "reducing from 2 million to 1 million."
Second, mechanism verification — letting phased results speak. Running data for a month or two is more persuasive than any debate.
Finally, it comes back to values themselves. If you've thoroughly explained the logic, data has verified it, and the other side still doesn't think it's right, can't sit down and rationally discuss something, and their understanding of value to users and to the Narwal brand still stays at "I just want to spend big money" — this is no longer cognitive misalignment, but a stance issue. In my eyes, they may not be fellow travelers.
You can't make everyone happy. As a founder, as CEO, the trap is wanting to make everyone happy — in the end, the only unhappy ones are yourself and the company. Often, alignment isn't about seeking common ground, but choosing fellow travelers.
Yingke: Does your understanding of human nature change?
Zhang Junbin: (Silence) I think of two phrases: "Of all virtues, filial piety is first — judge the heart, not the deed; of all evils, lust is first — judge the deed, not the heart."
I used to have too high a tolerance for certain actions. Now my judgment logic is: once there's error behavior that crosses a red line, there must be major action. Because following the principle of "judge the deed, not the heart" — the action itself is the answer; you don't need to further question the intention. Once such actions occur, it means this person is no longer suitable to remain in this position. But before, I might have had higher tolerance for these actions and just let them pass.
My sensitivity to people now, or perception of negative signals, is quite high. The reason is that these things are too damaging to organization and business.
But returning to attitude toward excellent partners, my perception of the good side of human nature remains — I don't completely deny the good side because of these things. It's just that for the dark side, the handling methods and understanding are different from before.
Yingke: Is there a way, through a mature organizational structure, to relatively level all the planks, while preserving the nice parts of human nature in this process?
Zhang Junbin: I definitely still have nice parts — it's not like I'm a demon now (laughs). It's just that before, you might see something was indeed done quite poorly, and verbally you'd say "OK?" There was even a phrase that circulated internally: "When Brother Bin says 'not bad,'" often your work was actually terrible. This was essentially unprincipled tolerance. But now I find that even being stricter, people don't think there's a problem. Of course this doesn't mean casually losing your temper — strictness doesn't equal being emotional.
Yingke: How do you want users to understand, "What kind of company is Narwal"?
Zhang Junbin: Narwal equals great products. Our positioning is global AI smart cleaning expert, doing product innovation centered on the household. Our approach and pace overseas differ from others.
At the end of last year, the team internally aligned on several key strategic brand values: emphasizing product innovation, emphasizing humanistic interaction between machines and people and space, truly helping users free their hands through smart products, and exploring more valuable things. When positioning is clear, external messaging becomes consistent.
Recently, our newly released overseas premium product line hit number one on the regional market new product ranking on its North American launch day. This achievement, in turn, confirms that we've finally clarified "who Narwal is" and understand how to communicate this in ways users can understand — rather than spending huge sums while users remain confused about who Narwal is and why they should buy.
Narwal Xiaoyao 003 (Source: Company)
Yingke: How do you understand users?
Zhang Junbin: We spend a lot of time on user research.
By the end of 2025, the team conducted over 5,000 in-depth user surveys domestically, including over 600 users reached by phone, each conversation lasting 30-40 minutes. We were pleasantly surprised to find that after all these years, users still see Narwal as an innovative brand — best mopping technology, obsessive about design and aesthetics, most disciplined product release rhythm and naming.
What users actually need isn't much, and it's pretty simple. Sometimes companies keep shouting "customer first," but if you're not even talking to your customers, how can you put them first? That's what we're pushing internally now: when you can't figure something out, go talk to users directly. Cut back on internal guesswork.
Yingke: Has Narwal's product strategy shifted? What's your plan for the low-end market?
Zhang Junbin: Our internal data shows that within our product mix, Narwal's mid-to-high-end models account for 90% of sales volume. For most competitors, that ratio is roughly 60% to 70%.
We want to go deep on product strength in the mid-to-high-end segment and push user experience to the extreme. But that doesn't mean we'll only do mid-to-high-end products. Narwal has coverage across high, mid, and low tiers. Ultra-low-end is off the table for us, but the entry-level low-end market is something we definitely need to pursue — from a global perspective, this segment still hasn't been effectively served.
The key to making low-end products isn't making them cheap. It's about subtraction — deciding what to cut and what to keep. That's the hardest part of value trade-offs.
Users buying entry-level products equally deserve access to core foundational features. This price band represents massive market space, accounting for 40% to 50% of industry shipment volume, yet it's long lacked genuinely thoughtful product offerings. This is also where Narwal will focus heavily in the second half of this year, with a new product launch. It won't be promoted as a flagship, but we want users in this price band to get the best experience available at that level.
Yingke: How do you assess the foundational base in overseas markets?
Zhang Junbin: The overseas market opportunity is enormous. Narwal recently brought on a seasoned DTC leader. I asked him, "Is the space big?" He said, "Massive. So much hasn't been done yet." What one of our teams has spent a year building looks like early-stage work to a professional.
Two data points illustrate this. First, overseas there's a large installed base of single-dustbin robot vacuums left behind by iRobot. An internal European user survey we conducted in March showed that 32% of users upgraded to Narwal from older models, citing inadequate mopping, sweeping, and obstacle avoidance on their previous machines. Second, in Europe and America, 50% to 60% of robot vacuums sold online are priced under $500 — a segment Narwal has barely touched.
After completing in-depth engagement across eight global cities last year, we also found that channels and users generally appreciate a brand that focuses on a single category and goes deep, rather than blindly chasing scale.
Narwal's current priority remains making great products. At every launch event or channel meeting, people ask: others have done yard work, lawn mowing, pool cleaning — when will you? We don't rule out entering these areas in the future, but there's still so much unsolved at ground level, at that 8-centimeter height.
Who has actually solved extreme edge cleaning? Who has really figured out table-leg navigation? No one. Too many experience gaps remain in between.
Yingke: You filter for value alignment when hiring. When facing the market, is that also a process of filtering users?
Zhang Junbin: Yes, but the vast majority of users aren't as complicated as people imagine.
I once talked at length with the founder of a well-known premium home appliance brand. He said, "Junbin, your marketing is a bit weak." (Laughs.) He shared a story. Early on they launched direct-drive washing machine technology, which added 100 yuan in cost, and the product just wouldn't sell. Then he came up with a solution: stand a coin on top of a running washing machine.
Belt-driven machines vibrate too much — the coin falls. But their direct-drive machine had minimal vibration, and the coin stood steady. That coin caused their washing machine sales to explode. He asked me: "Have you found your coin?"
Yingke: Do you think you've found it?
Zhang Junbin: Getting close. At least we're trying, and we're already seeing some verifiable results.
Many people probably haven't genuinely tried to find that "coin." The core reason they can't find it is insufficient depth of user understanding. You build so many features — which ones are the pain points users actually care about? Expression matters too. If you don't communicate directly enough, even the best technology won't travel. Telling users "open the machine and look — our technology is better" — they don't understand that at all.
You'll notice some companies with strong-looking technology and loud marketing end up selling terribly. The problem usually isn't the product experience itself, but a broken transmission chain.
So the key is finding that coin. Every SKU, every category needs its own coin. Through user conversations, we found that globally, when people mention Narwal, their first association is "exceptional mopping." So we should amplify that cognitive anchor — how do you prove you mop better? That's what we need to build. That coin is the moment that makes users believe you mop better.
Narwal's multi-SKU matrix (Image source: Company)
Yingke: Driving home the "AI-powered smart cleaning" brand perception
Yingke: There's a view that truly admirable companies aren't judged by scale or revenue, but by whether they're doing things with "compounding returns." Many companies switch categories too fast to build any compounding — four years on one product, then starting from scratch in year five. Over ten years, what has Narwal done that compounds?
Zhang Junbin: The way I understand compounding is like continuously watering a sapling — it keeps growing. Rather than waiting for flowers to bloom fully, wither, and then starting everything over.
Brand is 100% compounding. Narwal has been in the market for six years now. Our new premium product hitting #1 in North America and staying there for over a month — that's brand compounding paying off.
Second, channels compound. When you go sufficiently deep and broad on global channels, new products entering new channels and markets not only give you negotiating leverage but let you move in and out faster.
Third, technology compounds. In robot vacuums, traditional home appliance giants struggle to enter easily because the technology is still in its compounding phase — sustained investment creates positive accumulation. Some companies invest for a year or two, see no results, disband the team, and hire a new one to start over. Narwal doesn't operate that way. Our R&D investment is continuous and long-term.
Yingke: How do you determine whether Narwal's strong North American sales are brand-driven versus category-driven? How do you shift users from category-based decisions to brand-based decisions?
Zhang Junbin: A brand grows continuously, and that growth allows it to accommodate more category associations. Initially, Narwal meant robot vacuum-mop combos; stage two was becoming a smart home cleaning company; stage three might reach a new height, such as Embodied Artificial Intelligence.
Users fundamentally follow category decision logic — buying air conditioners, you first think of Gree or Midea. But if a brand is strong enough, like if Apple made air conditioners, users would definitely pay attention. Not because they assume the functionality is great, but because they believe Apple always delivers era-defining products.
As a brand grows, it necessarily moves from functional recognition to industry differentiation to emotional loyalty. Narwal is currently in the industry differentiation building phase. We want users to remember "AI-powered smart cleaning." If embodied intelligence develops and Narwal can actually build a smart caregiver robot, brand perception will leap forward again.
But if you're impatient and stuff in categories before your brand can accommodate them, you get backlash. The original positioning gets diluted, and the new categories don't stick either.
Yingke: Whether a company keeps getting better depends on being very firm about "being yourself." Some companies end up having to "be someone else."
Zhang Junbin: Narwal's hardest challenge is precisely that we're walking our own path. Finding a benchmark to copy would actually be easier, but that path doesn't go far. Internally, this is what we battle over most — their specs hit several thousand pascals, do you follow? Their hot water reaches 100°C, do you match?
The clearer you are about your own thing, the less easily you're swayed by outside noise. Ultimately, I think it comes back to self-consistency.
Yingke: What will you absolutely not follow, and what might you?
Zhang Junbin: Anything with genuinely zero user value, we won't follow.
We map four quadrants. Vertical axis is user value, horizontal axis is visible selling point.
High user value plus high visible selling point — we definitely follow. High user value but low visible selling point — we also do this, and it's where Narwal currently invests most. Neither, obviously we don't do.
The four quadrants Zhang Junbin referenced in conversation (Image source: Yingke)
Historically we've invested most in the upper-left quadrant — high user value, low visible selling point. Outsiders wonder why Narwal hasn't followed certain trends. But in reality, we've done reasonably well on both one and two.
Of course, Narwal's R&D and product development systems don't evaluate based on visible versus invisible. We primarily use user value as our metric.
Yingke: Have there been features you later thought "we could have done that then"?
Zhang Junbin: Basically no. If we didn't do something, it was because the technology wasn't fully ready at that moment.
Features with both user value and visible selling points — Narwal will do them, possibly with some delay, because internally we optimize until they're in the best state before launching.
Take mop extension, for example. It's a visible selling point with user value. Our solution launched six months later than industry alternatives, but performed better, so sales during launch period and promotional events were quite strong. Many extension solutions lack sufficient downward pressure on the mop, resulting in inadequate cleaning power; Narwal's solution maintains strong cleaning force even when extended.
Or the mid-roller brush hair tangling issue. Early competitors mostly used scissor-hand solutions; we later launched a single rotary arm solution that became a hit immediately upon release.
Yingke: Narwal previously mentioned plans in Embodied Artificial Intelligence. What's the current status of that project?
Zhang Junbin: We've been continuously exploring Embodied Artificial Intelligence, advancing it as a new product form with the goal of productization, not valuation.
Specifically, we have a dedicated team working on spatial organization and spatial cleaning. Currently, we've made decent progress on立体 surfaces like toilet cleaning and vertical tile cleaning. We've also made good headway on grasping and organizing multiple object types. Meanwhile, we're self-developing core components.
Honestly though, true commercialization still needs some time. Perhaps in two years, there will be a product with a robotic arm that can solve problems robot vacuums can't, truly entering households.
Of course, for Narwal, the current priority is focusing on the "AI-powered smart cleaning expert" brand perception and going deep on product layout. Embodied Artificial Intelligence is the future, but every step right now needs to be solid.
Yingke: What are the key elements to driving that home?
Zhang Junbin: Product. With our recent vacuum cleaner launch, almost every user who tried it gave positive feedback. But we also have a weakness — we haven't concentrated our firepower on the "narrative"穿透力 [penetrating power].
Narwal changed its slogan this year to "Enjoy a Freely Clean Home." I finalized this with the team remotely over WeChat while I was in Japan late last year. Our previous slogan was "Live Freely, Love Deeply," which was also a good one.
Yingke: Why the change?
Zhang Junbin: Because in the globalization process, I realized that brands need extreme focus. Products need their "coin" [core differentiator], and brand communication also needs to find its own "coin."
"Live Freely, Love Deeply" doesn't let users quickly understand: what does Narwal actually do? Nike could use this slogan too; any category could.
When a company's revenue hasn't reached tens of billions, a vague slogan creates excessively high cognitive costs. As a crucial medium for conveying content, it appears almost everywhere: manuals, product detail pages, packaging, even on the machines themselves.
Narwal's corporate development history brand wall (Source/Company)
Nike's earliest positioning was "building the highest-performance running shoes for serious runners" — very direct, not "Just Do It" from the start. All large enterprises' slogans follow an evolutionary path, and their changes reflect a brand's growth trajectory in a sense.
When a brand reaches maturity, rationality and emotionality coexist as dual elements, rising in alternating spirals. If there's only rationality, products become cold technology and users lack emotional connection; if there's only emotionality, the brand becomes hollow sentiment and users lack a foundation of trust.
At this stage, we've ultimately converged our brand positioning to "Global AI Smart Cleaning Expert," so users immediately know Narwal is a cleaning expert that can deliver better product experiences. But this isn't the endpoint. When the brand matures further, the slogan will gradually incorporate more emotional elements, carrying broader mindshare — this happens step by step.
Yingke: Is valuation a source of pressure for Narwal? The company did raise quite a bit of funding previously, and you mentioned continuing to push toward IPO. Market rumors have valuation dropping from over 30 billion to over 10 billion — that's a significant gap.
Zhang Junbin: We're actually fine on this front. We're not hung up on valuation. Valuation itself is dynamic now. The 30 billion valuation back then was indeed on the high side — the primary market was extremely hot, giving many companies valuations beyond reasonable ranges.
When the capital winter arrived, most companies returned to relatively rational valuation levels. This isn't a Narwal-specific problem; it's industry-wide. The real pressure is on many companies that can't even get their valuations to come down.
Valuation is floating; ultimately it depends on what kind of report card you can deliver post-IPO and over the next one to two years.
Narwal's current valuation is just a snapshot in time, not representative of the future. What's more important right now is running the business well, making good products, and pushing the company toward better scale and operational health — rather than obsessing over valuation itself. Shareholders haven't pressured me; instead, this lets me focus more on long-term, sustainable matters.
For me, IPO isn't an endpoint but a new starting point. Internally, we view it as a "second startup" — hoping that by that day, our organization, talent, and business will all be ready.
Yingke: What aspects of Narwal most reflect your personal characteristics?
Zhang Junbin: Everyone's obsession with product. Many people who join Narwal say your product requirements are too high. Because I have high standards for product, the entire team has high standards too.
We believe product is the "1" in everything, with all the "0"s attached to it. This understanding is aligned from top to bottom, and it's the fundamental reason Narwal has been able to continuously innovate over the years — and why these innovations have been accepted by the industry and users.
At the same time, Narwal is very restrained. Some companies will simultaneously open dozens of pre-research directions, but we won't. Identify one thing and polish it to perfection — this is also the compounding effect mentioned earlier, solving problems more thoroughly in one go. That's another characteristic.
Each founder's inner world largely determines the underlying logic and choice boundaries of their enterprise.
Yingke: In these ten years, has there been a moment when you felt your personal color actually constrained the company's forward progress?
Zhang Junbin: Too many... (pause). I'm a very nice person, which in business decisions often means too much emotion and not enough rationality. So I've had problems with people selection and judgment. This issue was already destined the moment Narwal expanded from 200 to 1,000 people and moved to Shenzhen in 2021 — this tribulation I had to cross myself.
During rapid expansion, we brought in a batch of executives who weren't necessarily suitable for Narwal. The root cause was my lack of mature judgment in selecting executives at that time, while also exposing major problems in managing executives. I'm often told I'm "not ruthless enough."
Looking back now, it was precisely these交错 [intertwined] tuition fees that made me truly begin examining my management values.
Yingke: What kind of person is a "ruthless" one? And what kind is someone less ruthless?
Zhang Junbin: A "ruthless" person makes decisions more crisply, cutting cleanly with hand raised. A less ruthless person thinks: is everyone happy? Will they mind? Some things you clearly know are wrong, but still let them develop. This is the aspect of my personality that has harmed me most.
Yingke: Is this "not ruthless enough" part hidden away, or have you stripped it off?
Zhang Junbin: Hiding and stripping are two different things. I think I've stripped off a large portion; a small remaining part is hidden away. Stripping at the foundation of personality takes time.
Yingke: Could that un-stripped small portion affect your decisions in the future?
Zhang Junbin: Unlikely. My tolerance window is much narrower than before. Or rather, my patience threshold for seeing results presented has lowered significantly.
I need to see results quickly, and I'll dive deep into details to verify whether things are truly done right. If details are done right but there's no result, I'll accept that — the difficulty exceeded expectations. But if details aren't done well and there's no result, that's an execution problem.
Before, I considered people's feelings more. My tolerance for results was relatively high. Someone might not be suitable for a role, achieving sixty or seventy percent, and I'd accept it. Now, sorry, I can't accept that.
To give an example, my requirements for presentation quality are much higher than in the past. Because presentation quality itself is the externalization of a person's depth of thinking.
Yingke: Do you prefer your former self or your current self?
Zhang Junbin: I later realized a problem: overly tolerant managers actually deprive teams of room to grow. If you think everything is fine, people can't raise their own standards.
Some people perform exceptionally well, others less so. Those who perform exceptionally well are self-driven and don't need you to tell them; but when they see those who perform less well never being called out on problems, and you don't mind, this itself harms the excellent performers. Those who perform less well, lacking self-requirements and real feedback, naturally can't grow; they don't know they're making very basic mistakes, because after making mistakes they discover you can accept it. This harms everyone, and harms the company.
A company with extremely long longboards and extremely short shortboards has problems. The bucket effect will inevitably leak.
Yingke: There are roughly several types of leaders — Steve Jobs and Elon Musk are one kind, Microsoft's Satya Nadella and Tim Cook are another. Which did you lean toward before? Which are you now, or hope to become?
Zhang Junbin: I probably leaned more toward a "nice guy" style of manager before. Cared too much about team feelings, unwilling to confront conflict. Essentially a passive tolerance.
In the future, I hope to become a relatively steady, less impatient person. Rational, calm, but with clear bottom lines — managing this company in this state. That very aggressive, all-in gambling personality doesn't seem to suit me.
Yingke: We've interviewed companies before that, in later stages, bring in professional managers and let founders cash out and exit. Have you considered this option?
Zhang Junbin: The core question to answer about cashing out is: after doing this, what do you do next?
Narwal to me is like my own child. Cashing out is essentially selling your child, right. If you believe this child still has huge room to grow, then continuing to accompany them as they grow is itself a very happy thing.
Every founder has their own psychological safety line, with different bottom lines. Once I've crossed that safety line, what truly needs thinking about is: what do I actually want to do with the rest of my life?
If I can build Narwal into a world-class brand, that's also quite interesting. Some people have great desire — maybe they need to sell a company to satisfy it. Some may feel greater accomplishment in doing this enterprise well.
People live, they always need to do something.
Yingke: In today's business environment, wouldn't being a bit more "fierce" sometimes make you stand out more and leverage greater resources?
Zhang Junbin: Rationality isn't not gambling — it's gambling very precisely. All big bets depend on you seeing absolute business opportunity.
What I mean by rational and calm is being able to reasonably assess risk; "irrational" is doing all-in without assessing risk at all. This doesn't mean we don't do all-in, but rather that we may do more thorough, more rational judgment and prediction of risk before choosing all-in.
Timing is very important.
Yingke: On the cleaning front, in the next 5 to 10 years, do you see a very clear timing?
Zhang Junbin: Yes. The moment the inflection point of penetration rate arrives — that's when you should bet big.
Yingke: Why hasn't it arrived after all these years?
Zhang Junbin: First, there's still a gap in product experience. Narwal achieving the highest好评率 [positive review rate] nationwide was selected from 420,000 reviews, but I still frequently criticize employees responsible for quality and production. This precisely reflects that the industry's overall maturity is far from adequate. Robot vacuums are extremely complex — compared to ordinary home appliances, they have dozens of motors; one breaks and the whole machine has problems. This is essentially the矛盾 [contradiction] between product complexity and reliability.
I believe there will come a day when robot vacuums can work for three or four years without breaking, continuously operating, while costs decline and intelligence levels improve — that's when the inflection point arrives. All home appliances go through this stage; for example, Narwal invented water hookup and auto mop washing, and when users universally accept such functions, penetration will surge steeply.
The key is to answer a fundamental question: in the future, is floor cleaning something humans should be doing? If not, then the inflection point will inevitably come.
In fact, change is already happening. When Tencent invested in us last year, they conducted a survey of 10,000 people, having done the same five years ago. Data shows repurchase rates rose from sixty to seventy percent five years ago to nearly ninety percent now. As product experience improves, many people already treat robot vacuums as necessities.
But there are still large numbers of users hurt by early inferior products — they naturally view this category as a toy. Those products not only damaged their own brands but also透支 [overdrew] the entire industry's trust, which takes time to repair.
Right now, the products that truly meet user needs still carry relatively high price tags, so the market hasn't fully opened up yet. On top of that, in the cutthroat competition, many companies have loaded up on features that add no real value for users — only creating more quality issues and complexity. So the key going forward is figuring out how to get more users to give a thumbs-up, boost NPS, and drive word-of-mouth.
The inflection point will come, but until then, Narwal needs to tell its own story well.
