Back when he started out, Zhang Junbin spent three years building his first robotic vacuum cleaner, secretly hoping that selling 10,000 units would be a win. That little machine ended up shifting hundreds of thousands of units and turned him into a hardware star everyone looked up to. Fast forward a decade, and Narwal has scored over 10 rounds of funding, becoming a unicorn worth more than ten billion yuan.
Today, Narwal has nearly 2,000 employees, with three major business units: robot vacuums, handheld cleaners, and embodied intelligence. Its products are sold in over 30 countries. But with growth comes pressure: fierce competition in the robot vacuum space, ongoing internal organizational shifts, and a whole new set of questions about how to pick the right people.
Zhang decided now is the time to personally take charge of global marketing—an area he’s never really been comfortable with. He used to think of himself as a “super product manager,” spending all his energy on engineering and product, barely paying attention to marketing. But by the second half of 2025, he realized his thinking was wrong: product and marketing aren’t additive—they’re multiplicative. If either one is near zero, the result is zero.
What the outside world really wants to know is something else: the competitive landscape is being rewritten fast—Roborock and Dreame are battling for the high end, DJI is jumping in from outside the industry. Can Narwal hold its ground?
Executive changes have also grabbed attention. Several high-level departures have fueled rumors of strategic disagreements and internal turmoil. In the interview, Zhang admitted his hiring criteria have undergone a fundamental shift. “I used to prioritize capability. Now I put values ahead of skills. There’s a type of person who’s incredibly capable and can deliver short-term fireworks, but in the long run they do huge damage. I absolutely won’t hire that kind anymore.”
He also said Narwal’s major restructuring is mostly done, and the gaps between marketing and product are being plugged one by one. As a result, the group’s gross margin has climbed by more than ten percentage points. By the end of Q3 this year, organizational efficiency should be in a good enough shape—that might be the moment Narwal is ready to fire.
On Narwal’s tenth anniversary, Zhang sat down with media including Interface News to address nearly every hot topic. Here’s the edited transcript:
On the Industry
Q: Narwal’s online sales of robot vacuums in Q1 this year declined. How do you see the overall industry?
Zhang: In the first half of last year, national subsidies provided a massive boost. In 2025, that ripple effect still lingered, so most of the growth came from those subsidies. That’s one thing. Now this year, although online sales dipped a bit in Q1, offline growth held steady. After the subsidies faded, offline channels started recovering faster than expected. In the first four months of this year, Narwal’s offline sales grew over 150%. Down the road, the cleaning industry’s offline footprint will keep expanding.
Q: Could the subsidies backfire on current demand? Where is the market momentum this year?
Zhang: Traditional home appliances saw a pretty sharp drop in Q1—things like air conditioners, refrigerators, washing machines fell quite a bit. But the cleaning category barely dipped, almost flat. That’s compared to the crazy spike in Q1 2025 driven by subsidies. So actually, Q1 this year wasn’t bad—given there were no subsidies.
Looking long-term, the cleaning industry—especially premium categories like floor washers, high-end vacuums, mite removers, and robot vacuums—will keep getting smarter. There’s a lot of unmet demand that’s now being addressed through product upgrades, including deeper penetration into lower-tier cities. That’s the engine for the next two to three years.
Q: Has the market penetration for robot vacuums met your expectations over the past year? When will the inflection point come?
Zhang: Penetration is still far from where it should be in any region. The reasons are product quality, user experience, and price—these limit how fast the category can spread. Think about washing machines 20 years ago—nobody expected them to have such high adoption. It took 20 years of upgrades to get to today’s experience. Robot vacuums still have room to improve, and plenty of it. That’s point one.
Second, the supply chain for robot vacuums is maturing, but volumes need to ramp up gradually so the chain can scale up for mass production. That takes time.
Third, as everyone keeps piling on features, quality actually goes down. You need time to digest all those new additions. When quality, cost, and experience hit a sweet spot, that’s probably when the inflection point arrives. And I personally think it’ll happen within three years.
Q: Adding new features lowers overall quality?
Zhang: Because they’re not mature enough.
Q: For example?
Zhang: Obstacle crossing, for instance. A lot of users get stuck on thresholds and stuff. The designs we’ve seen so far aren’t really mature, so we’re not rushing to release ours. But it’s coming soon.
Q: You mentioned scale brings down cost. What kind of volume would trigger a qualitative shift?
Zhang: For the domestic market, at least 50 million units a year.
Q: Brands like DJI, which have had success overseas, are now entering robot vacuums. Narwal is also going global. Does that make you anxious? What is Narwal’s edge in the cleaning space?
Zhang: Not really. I was a bit worried at first. But I think DJI actually did the industry a big favor—they expanded the audience and brought in a lot of geeks. The key question whenever a new competitor enters: are they taking a slice of your existing pie, or growing the whole pie? DJI is the latter. They segment their own fan base, and that conversion is high.
Q: Weren’t those people considering robot vacuums before?
Zhang: Very few. Plus, DJI’s global brand awareness gets more people paying attention to robot vacuums overall. Home appliance buyers are mostly white-collar workers and moms, but the tech and geek crowd? We haven’t reached that many, because they’re not our core users.
Q: But those folks might be harder to convert to other brands.
Zhang: Not necessarily.
Look at it this way: when DJI enters this space, people who are loyal to DJI might discover there are more specialized cleaning brands out there. They now have options. Also, brands that do well in this industry need solid brand equity, years of experience, user data, and continuous operation—new entrants can’t overturn that overnight. But we do have to keep investing to hold our ground.
Second, robot vacuums and cleaning products are core to Narwal, whereas for DJI it’s just one line. We’ve got nearly 2,000 people all-in on the cleaning industry. The user experience we deliver speaks for itself. We stay focused, which is why we’re deliberate and why we stick to AI cleaning. That’s our strategy. When users make buying decisions, they think category first, then brand. For air conditioners, you think Gree or Midea, not other brands. That kind of brand positioning grows as the brand matures.
Q: iRobot was acquired by a Chinese supplier last year and is starting to push in China. Does that affect the market?
Zhang: Not much. But the global robot vacuum market is definitely heading toward full Chinese manufacturing.
On Business Operations
Q: Why didn’t you attend AWE this year?
Zhang: Big expos are usually for two reasons: either you’re entering a brand-new market and need to meet media and channel partners at scale, or you have a game-changing product to launch and want that concentrated buzz. For Narwal right now, we’d rather put resources into deep local operations—go deep, go solid, and keep talking to users.
Q: Narwal used to be slower than the industry. In the last couple of years, you’ve accelerated. Was that a conscious choice or forced? Which pace do you prefer?
Zhang: It’s a bit of both. It’s not purely competitors forcing us to churn out more products. It’s that I’ve come to understand that different price bands need different products. In the past, Narwal launched one model a year, but that couldn’t satisfy everyone. Whether high-end, mid-range, or entry-level, the core is still delivering a great user experience. So we’re building SKUs around that idea. Think BMW: 3 Series, 5 Series, 7 Series. You need them all.
But you can’t go crazy with SKUs either. Too many means spreading R&D thin, and you can’t polish anything. It’s a balancing act. I think you should have offerings in three price bands, and at least one product a year that really clicks with users.
Compared to 2024, we actually added more SKUs in 2025—though nobody knows that. We put out a bunch around robot vacuums and floor washers, but we found that more SKUs didn’t bring proportional gains. So this year we adjusted: focus R&D, refine the best products.
Q: You’re so cautious about SKU expansion. Is the pressure technical or financial?
Zhang: It’s mainly about senior management bandwidth. Money isn’t the issue. But if your energy can’t cover it, you risk going off track. Maybe you hire the wrong manager, and you’d be better off not rushing. That doesn’t mean we’re not looking. Choosing managers is a long-term thing; you’ve got to pick really carefully. That’s the first and most important step.
Q: But more SKUs didn’t seem to boost sales much?
Zhang: Because with more SKUs, your marketing gets unfocused too. Nobody is systematically organizing the value chain and market plan. So now we’re sitting down together more often to review these things. Basic SKUs get upgraded—maybe two in H1, two in H2, plus one entry-level model. That’s the rule. On top of those five, how many derivative SKUs to add? That’s what we need to figure out.
Q: Narwal now has not just robot vacuums but floor washers, vacuum cleaners, mite removers. What drives this multi-category expansion?
Zhang: Our brand positioning is “Global AI Smart Cleaning Expert.” We’ll keep expanding around AI and cleaning, replicating the product innovation we did with robot vacuums and scaling up revenue. That’s the unified direction for brand expansion.
Take vacuum cleaners. The market is huge, and there are still plenty of pain points that haven’t been solved. I personally think vacuum cleaners are due for a second wave. Hard to predict volumes, but there’s still big room for improvement in user experience, maintenance-free features, and cleaning power.
Right now we have three business units: handheld cleaning, robot vacuums, and embodied intelligence. We’re mainly focused on these three, and also planning a new unit for early next year.
Q: Narwal is positioned as premium, but you mentioned mid-range and entry-level have big markets too. How do you choose between scale and profit?
Zhang: Until I’ve optimized the marketing organization’s efficiency to a point I’m happy with, I won’t chase scale. Throwing money at it would just be flushing it away. At this stage, I need to focus on improving organizational efficiency. Right now, profit comes first. Once we reach a certain point, I’ll switch to scale priority.
Q: Is this a new insight since you started leading marketing?
Zhang: After I took over marketing, I saw so many efficiency gaps. Now we’ve patched about 80% of them. These aren’t just marketing team gaps—they’re gaps in how marketing and product work together. The overall optimization on both sides wasn’t great.
For example, the ordering process: from front-end demand to back-end production, we needed to shorten communication time. It used to be long, and we’d finish production only to find the front-end no longer wanted it. Just fixing that one thing dramatically improved inventory turnover.
Q: What’s Narwal’s overseas revenue share now? Where will you break through next?
Zhang: From January to April 2026, overseas revenue was about 40% of total. Group revenue grew slightly.
On overseas markets: first, product. We used to only focus on Chinese user needs, but now we place high importance on understanding overseas needs and usage scenarios. Most upcoming products will be designed primarily for overseas markets.
Second, from a business perspective, online space is still huge. Our B2C on Amazon hasn’t been ideal—still tons of room. Recently we launched the Xiaoyao 003 series (overseas version FLOW2), and with significantly improved marketing efficiency, it hit #1 in the new product rankings for robot vacuums on Amazon US.
Third, personnel training and development. We’ve brought in some external consultants to accelerate capability building. Finally, there’s IT infrastructure to ensure ad efficiency and better business progress.
Overall, the global robot vacuum market is still growing, and there are many countries we haven’t entered. Although Narwal’s overseas revenue is 40%, just a few countries account for 80% of that. There’s still plenty of room. And we’ve found that once we build a premium product, replicating it to other regions is fast—we don’t have to redo everything for every country. That could be the engine for our next phase of overseas expansion.
Our share in the mid-to-high price band is the highest—about 90%, while the industry average is 60-70%. But the bulk of overseas volume is still in entry-level products. The shift from standalone units to all-in-one stations is a huge opportunity for us. That’s another growth driver.
Q: What will be Narwal’s future growth engine?
I’ve always been focused on product and R&D. I only started managing global marketing around September or October last year. Before that, I paid very little attention to marketing. I used to think it was additive: “product + marketing.” If you build a great product, you can afford to neglect marketing a bit. But I later realized it’s actually multiplicative: “product × marketing.” That means both need to be excellent to take the company to the next level.
So I took over marketing, and I saw tons of opportunities. A big part of Narwal’s future momentum will come from my renewed focus on marketing—meaning global business layout and operational efficiency. I used to think this wasn’t a high priority, but now my thinking and focus have to evolve.
Since I stepped in, I’ve made a lot of changes: channel optimization, pricing systems, partner relationships, contract terms. There’s huge room here, and our gross margin has already increased by over ten percentage points.
I recently talked to a senior exec at a major appliance company. He said I need to find Narwal’s “coin.” Back in the day, Casarte’s washing machine introduced a direct-drive motor that made vibrations tiny. But it was hard to show as a selling point directly—so they placed a coin on it. On other machines, the coin would fall over from vibrations, but on Casarte it stayed upright for a long time.
So now I’m leading the team to figure out what Narwal’s real competitive edge is, and find the coin that tells that story. I’m putting a lot of emphasis on clearly expressing “what we have that others don’t, and why ours is better.” Some founders are great at that, but Narwal has been too conservative. That’s something we need to amplify going forward.
Q: How are you learning marketing? What does this shift mean for you?
Zhang: It’s definitely outside my comfort zone, but that doesn’t mean I can’t learn fast and iterate. Those are two different things. After taking over marketing, I’ve been learning like crazy. In October, I worked until 3 or 4 AM every day—intense learning, diving deep into the business. I met a ton of people, talked to at least three or four top-tier marketing experts in China, just exchanging ideas.
But marketing is different from tech. In tech, it’s engineering logic—clear right and wrong boundaries. In marketing, there are first principles too, but a lot more things that can’t be precisely quantified. Like hiring a KOL, picking a few campaign angles—who can guarantee the data will be great? A lot of marketing can’t be measured exactly.
But the basic systemic stuff can be learned fast. For example, setting up a clear financial system to monitor and adjust—that’s a mathematical optimization problem, based on deep business understanding. If your company has tons of products, each targeting different segments with different life cycles, resource allocation, and organizational capabilities—that’s essentially a math problem. I need to find the optimal marketing path for every product, talk to users, make sure they remember Narwal, and build brand awareness. R&D has clear boundaries, but marketing is more flexible. That requires deep involvement and fast learning.
Q: Have you summed up any lessons?
Zhang: One big advantage of being a founder is access to top talent. You have to be bold enough to learn from the best, with humility. That’s crucial. Once you have a direction, the next thing might be just hard work. You can’t just play golf all day and not work. I’m still grinding on the front line, working late every day.
Then it comes down to execution. Only after executing do you discover the real sticking points. Then you go ask others about those points, and you see very different perspectives. For example, why is Huawei’s overseas structure so solid? What’s the logic behind the ratio of Chinese vs. local employees, who leads, who supports, who are experts, who are managers? There’s a system behind it. After hearing it, you realize it makes a lot of sense. I’ve been facing some country-specific operational issues myself, so I’ve been talking to many people.
Q: You started a podcast a few years ago to build a personal brand. Has it made a dent in the brand?
Zhang: Hard to say. It definitely strengthened relationships with our partners—that’s been validated. Whether it directly impacts sales is tough to tell. Because of limited bandwidth, I stopped for about half a year, but I’ll pick it up again once I free up some energy.
Q: Where is your marketing budget concentrated now?
Zhang: Audience expansion and precision operations—getting more people to know and trust Narwal. On the brand side, we want to elevate Narwal’s image another notch. Early on, people thought of us as an internet-famous brand. Then gradually, as a decent mid-range brand. Now after releasing Xiaoyao, we want people to see Narwal as a premium brand. We have quite a few plans around this, including collaborations. That’s worthwhile.
On Talent
Q: The wave of senior departures at Narwal—are outsiders right to guess there were strategic clashes between you and them?
Zhang: Different executives left for different reasons. Most were amicable; no ugly breakups. What matters now is whether we can attract new top talent and what kind of people fit Narwal’s rhythm.
Sure, some left, but many excellent people are joining. The new team keeps surprising me, so I’m not too worried about the business. For new executives, I used to value capability above all. Now I still look at capability, but I put much more weight on values. After certain events, values have become a bigger factor. The current senior team is integrating really well—it feels smoother.
I always emphasize “unity from top to bottom.” That’s not just strategic alignment, but value alignment too. I’ve come to realize that how far a company can go depends not just on whether the direction is right, but whether the team is truly in sync.
Speaking of values, a lot of the good things Huawei does can’t be replicated in other companies—because the values are different. For example, Huawei’s “putting strivers first” and their strong rotation system—some companies try to copy it, but their employees leave as soon as they’re rotated, causing huge attrition. Many teachers say don’t blindly copy Huawei, because a lot of it isn’t learnable. It’s not about the system or incentives or policies—it’s about values.
If someone is super capable, efficient, and can deliver explosive short-term results, but in the long run they’re actually harmful—I won’t hire that person, period. On the other hand, if someone is capable, steady, maybe a bit slower at seizing opportunities, I’d actually choose the second type.
Q: There’s been a hardware startup wave in Shenzhen the last couple of years, and there’s talk of “Narwal alumni.” Do you feel pressure on talent retention?
Zhang: People who want to start their own thing are hard to keep. Many aren’t driven by bigger money—they have their own dreams and strong growth needs. The company can’t always offer them a corresponding position. A manager-level employee at Narwal might become a co-founder at their own startup. That kind of person will leave any company; they’ll take a pay cut to join a startup. It’s not about salary.
Not every field is hot right now for startups. AI-related ones are, others maybe not so much. I feel this wave is trending downward. There aren’t that many good projects, and good product ideas are probably already being worked on by decent companies. I keep an eye on the outside. Unless you have a truly groundbreaking product definition, fundraising won’t be as smooth as you’d expect.
So for those employees—if you can’t keep them, encourage them to go start something. I still have good relationships with them. I’ve tracked it: about 30% have failed, 50-60% are still alive, and one even became a unicorn.
On Technology
Q: In 2026, the hottest investment topics are AI and embodied intelligence. Even early-stage embodied startups with no market validation can raise 1 or 2 billion yuan in Series A. From multimodal models to hardware design to product direction, many robot vacuum companies claim they can pivot into embodied intelligence. Why hasn’t Narwal said that clearly?
Zhang: Not saying it publicly doesn’t mean we’re not doing it. Would slapping on the “embodied intelligence” label really add more value to users? I’m not so sure. On the product side, we’re very aggressive. But on the PR side, we haven’t called ourselves an embodied intelligence company.
Q: In last year’s interview, you mentioned moving from floor cleaning to space cleaning—like organizing and tidying up. It’s been over a year. How’s the R&D progress on space cleaning?
Zhang: We’ve always had an embodied team working on space organization and cleaning. In some ways, we’re pretty good at cleaning specific vertical surfaces—like toilets and vertical tiles. For grasping and storing various objects, we’re doing well too. We keep advancing in this direction and developing core components ourselves.
But truly commercializing this and getting it into millions of homes will take more time. There are a lot of algorithmic challenges, product definition challenges, and electromechanical system innovations involved. But as algorithms improve, data accumulates, and we bring in more talent, I feel that moment is getting closer. Personally, I think in two years we might have a product with a robotic arm that can solve a lot of things robot vacuums can’t.
Q: Last year you talked about breakthroughs in joints and vision. What other areas will you optimize and iterate on? We’re seeing flying robot vacuums, stair-climbing ones—but widespread adoption is still far off. In general, where is the technological increment?
Zhang: First, for ordinary users—like moms—can the human-machine interaction be simple enough? Can the failure rate be low enough? Those two are probably the most important.
We’ve been investing in technology all along, but on the product definition side, we’re thinking more deeply. In the past, we were mostly making tech reserves to see how high we could go. Now we’ve hit some ceilings, and we need to decide if we can turn them into products. That’s a very real question.
If the foundational capabilities aren’t built, you can’t make the product. Like multi-touch technology—if it hadn’t been sold to Apple, the first iPhone wouldn’t have had many features. So for this level of embodied product, you explore the tech ceiling first, then build the product.
Q: Many products get copied quickly by competitors. Does that mean technology isn’t as important anymore at this stage?
Zhang: It’s all important. For many features, Narwal has “what others have, but better.” If we can tell that story well, all our technology becomes valuable. The core issue is that we haven’t yet executed on that first part to the fullest.
Q: When will your embodied product go into mass production?
Zhang: We’ll keep it under wraps for now, but we will definitely have a complete product form.
On Growth
Q: Over the past decade, what’s the biggest change in you and the company since the early days?
Zhang: I’ve grown. From an engineer to a product manager, then to an organizational manager, then to a cross-domain manager, and now to thinking about strategy. At each stage, there were core contradictions in the company—but you might not identify them perfectly. And sometimes there are two or three core contradictions, but your capability is limited, so you can only tackle one. You have to iterate and grow fast through those contradictions.
In Narwal’s early days, from founding to launching the first product, I was more of a product manager and system engineer. I don’t think I was a mature manager back then. When we grew from 200 people in Shenzhen to 1,000, I hadn’t completed a good management transition. I was still a good product manager and system engineer. Managing a 1,000-person team was definitely tough.
Gradually, as my management skills improved, I realized I needed to understand other domains—beyond R&D, production, and product projects, I also needed to learn about service, logistics, finance, HR, legal, etc. That was a period of cross-domain knowledge growth and cross-domain team management. Going further, I had to think about strategy. Sometimes you only do strategic insight but not the rest. Or you do insight and execution but lack strategic alignment—so the top and bottom aren’t on the same page.
This year we focused on goal decomposition. Two business units held a three-day meeting to align on strategy, down to execution details. For each strategic goal, we clarified the three most important things for the department this year and how to get them done. In many companies, the strategy is one thing, but when teams set their own goals, they list a bunch of broad stuff with little connection to the strategy. We were like that too—the front-line didn’t know what the top wanted, and the top found the front-line too slow. So we’re upgrading our approach.
On values, I’ve also come to understand what kind of people fit Narwal. Values aren’t right or wrong, but some match better. Some people thrive at Narwal, some don’t. Often it’s not about capability—it’s values.
Q: How do you steer your company toward what you believe is the right goal?
Zhang: I aim for business success while delivering better products to users and improving their lives. I think people need to be self-consistent. If you chase only commercial success, you might have to do things against your will or feel miserable. I might not be able to make that feel consistent. Some people achieve massive commercial success but aren’t happy at all. Others might be poor but joyful every day.
Once you think it through, your choices become different. Being self-consistent doesn’t mean sacrificing commercial success. I believe any form of commercial success counts.
Narwal knows its positioning well today: become a global AI smart cleaning expert. Our business units and products are built around that. I often ask myself: what kind of person do I want to be? What kind of work truly makes me happy? Actually, the logic of thinking about people, companies, and even life itself is similar—we all need to find our own path and walk it with determination.
Q: Has Narwal achieved the ten-year goal you set at the beginning?
Zhang: I never had a ten-year plan. But at least we’re still alive. Back then, it was simple: I just wanted to make a good product. I really enjoyed the process of building it. Spending three years on the first product, feeling thrilled when it was done, and seeing it blow up. I thought if we could sell 10,000 units a year, that would be great. I never imagined we’d sell hundreds of thousands. But once we entered fierce competition, that phase was tough—I didn’t realize the world could be so unpredictable.
Q: Ten years from now, when people think of Narwal or Zhang Junbin, what do you want them to remember?
Zhang: Ten years from now, Narwal could be a world-class brand. People will think: if Narwal makes it, it must be top-notch. And it’ll be a truly global brand with a sales network spanning the world. That’s definitely the goal I hope we reach.
As for what others think of me—I don’t really care that much. Or rather, let me talk about how I see myself. I hope by then I’ll be very self-consistent. I’ll handle everything with ease, and maybe I can finally call myself an entrepreneur, not just a startup founder. There’s still a big gap, I think.
