I once wrote a very long work email about this topic. It had been years since I last wrote something that long. The reason was simple: the issue was not a single failed metric, but a chain of misunderstandings about market goals, operating logic, and product priorities.
The company had been talking about KPIs for years, yet it was only this month that I learned the actual requirement for the China market: 8% month-over-month growth. I had been working remotely for a long time and had limited communication with the U.S. team while I was in China, so that partly explains how this happened. But that is still not the real problem.
The real problem is that a KPI that depends on coordinated team effort was reduced to a percentage that frontline staff were expected to hit without the data needed to plan, execute, or evaluate anything.
We had no visibility into the most basic operating numbers for the web and app products: traffic sources, IP and pageview volume, order volume, churn rate, churn pages, conversion rate, and so on. We also lacked data on marketing performance: the effectiveness of campaigns, how much traffic each promotion brought in, and how many conversions it generated. Even the U.S. marketing team did not seem to have a clear picture. The company says China matters, yet the people working closest to the market were effectively asked to move forward blind. Under those conditions, how is a KPI supposed to be achieved?
Japan and Korea have much smaller populations than China, yet their results are several times better. Why? In China, a single e-commerce promotion can generate astonishing transaction volume in one day, yet we capture so little of that market. Why?
My view is that the company’s failure in China comes down to two things: a mistaken belief in globalization, and an operating model built around traffic.
The globalization illusion
There used to be a common assumption that major U.S. internet companies naturally build for the world, and that good products should therefore work everywhere. It is true that giants such as Google and Facebook still dominate much of the global internet outside a few exceptional markets. But if you look closely at rankings and market behavior, one fact is hard to ignore: U.S. internet companies no longer win by default in every market.
Their early global advantage had very specific causes.
First, the internet industry began in the United States, and that first-mover advantage was enormous. By the time U.S. platforms became highly mature and sophisticated, local internet ecosystems in many other countries were only just emerging. Those markets often lacked both capital and talent, so meaningful competition did not exist.
Second, many countries and regions simply did not have enough product and engineering talent to build serious internet products locally, especially in technically demanding categories such as search engines, large social networks, messaging platforms, and complex e-commerce systems. Under those conditions, local demand was suppressed. People used U.S. products not because they were the best fit, but because there were no real alternatives.
Third, many markets were too small to justify expensive localization. When development costs are fixed, it is more efficient to satisfy the needs of the largest markets first. That means the needs of smaller countries and regions are often ignored entirely.
Those factors explain why U.S. internet giants were once able to sweep across market after market. It was not because their products were universally suitable. It was because many countries had no choice.
And once real local options appear, the story changes.
In Singapore, the leading e-commerce platform has been Qoo10 rather than Amazon or eBay, and not even Lazada. In Southeast Asia, Grab became the leading ride-hailing platform rather than Uber. In Vietnam, the leading messaging app is Zalo. In Japan, it is Line. In the Middle East, Careem became the leading ride-hailing platform, and MumzWorld became a major mother-and-baby e-commerce player. In Africa, Kilimall rose in e-commerce. Some of these successes were even built by Chinese teams, which says a great deal by itself.
The globalization mistake leads companies to believe one product can conquer the world unchanged. Once you believe that, your global operating strategy often becomes traffic acquisition: get enough users in the door and the market will follow. That may have worked in some places. It does not explain China.
The reason a U.S. company can succeed in Japan or Korea but struggle in China is not mysterious. China has its own product, engineering, and platform capabilities. It can build for its own users and satisfy its own needs. When a market can create its own solutions, it will make its own choices.
Many U.S. companies still do not truly understand China. Over the past decade, China’s internet industry did not merely catch up; in many respects it overtook everyone on the curve. In some areas, the U.S. is behind without realizing it. Mobile payments are an obvious example. WeChat Pay and Alipay effectively allowed China to leap past the credit-card stage that still shapes consumer behavior in many Western markets.
In China, nearly every part of daily life—food, clothing, housing, transport, errands, payments, services—has been deeply integrated into internet products. With one phone, a person can handle the vast majority of everyday needs. In many cases, the speed and convenience of those services are beyond what most Americans would expect; the best experiences are measured in seconds.
If a company approaches China as though it were just another overseas traffic market, it is already starting from the wrong premise.
The limits of traffic thinking
Whether through partnerships or paid promotion, the company’s operating logic has long been traffic-oriented. That means decisions get evaluated through a traffic lens, and market operators are treated mainly as people who push volume. But China’s internet environment has been changing for years. Since around 2016, the era of traffic-first growth has been steadily fading.
Ten years ago, users had far less influence over how products spread. If a company occupied important channels and entry points, and did a decent job with conversion and distribution, the numbers could look great.
That is no longer enough.
Today, even if you hold major traffic channels and important distribution positions, growth becomes more expensive and less effective if users do not endorse the product. If large numbers of real people are not willing to say the product is good, traffic conversion falls, acquisition costs rise, and retention weakens.
In simple terms, broad, channel-led traffic strategies have become inefficient. A few years ago, the cost of acquiring a user might have been around 3 yuan. Now it can easily be more than ten yuan.
The rise of social media has fundamentally changed the logic of information distribution. Information is now more distributed and decentralized. In the past, if a major public event happened, people largely relied on news media and search behavior to learn about it. Now information reaches users immediately through platforms such as Weibo, private social feeds, and knowledge communities. The way people discover, trust, and discuss things has changed.
That has major implications for product operations. The real foundation now is not simply whether you can buy traffic, but whether people are willing to speak for your product on social platforms and in their own circles. That endorsement is the base layer. Once it exists, promotion and conversion become far more efficient. Without it, many efforts are wasted.
Cross-border shopping has always had a high barrier to entry. It requires learning, comparison, and a willingness to deal with unfamiliar overseas products and systems. That uncertainty makes decision-making harder for users. In our case, the user reward mechanism encouraged users to talk about us, and that public sharing acted as social proof. Community operations also created stronger ties among users. Before the .cn shopping site was launched, the community had already become the second-largest traffic source to the .com site after Google AdWords. That says a lot about how users actually made purchase decisions.
In the past, operators acted mainly as drivers of growth. Now they have to be catalysts. They need to move users emotionally, while also staying responsive to users themselves. That means shifting from traffic orientation to user orientation: building stronger emotional connections with users, encouraging stronger user-to-user relationships, and maintaining more frequent interaction.
In the early stages of building the community, we experienced several shipping crises. Each time, I wrote extensively to communicate with users and created more opportunities for users to interact with each other. That emotional connection mattered. It kept people from feeling isolated, confused, or suspicious, and helped prevent frustration from turning into blame.
How e-commerce changed in China
The internet has been going through another major transition, often described as consumption upgrading. This applies both to content consumption and to product consumption. In retail, one of the clearest outcomes has been the rise of content commerce.
Traditional e-commerce is transactional. Its main way of driving sales is promotion: flash sales, full-reduction discounts, gifts with purchase, markdowns, mix-and-match offers, and other price-led incentives. In the early years of e-commerce, this model worked very well. Many platforms used it to expand SKU count, strengthen supply chains, and scale their businesses rapidly. But after the high-growth phase, this model entered a slower or even declining stage. The most obvious examples are the large, established Chinese platforms.
The weaknesses are obvious. Constant discounting erodes margins and makes profitability difficult. For users, too many products become tiring. Constant price fluctuations create distrust. Users get trained to compare endlessly and become less loyal. Over time, they also become numb to promotional tactics, and purchase motivation falls.
That environment gave rise to content-led shopping guides. One well-known example was Mogujie, which evolved into a social content e-commerce platform and achieved real success.
Later, as cross-border e-commerce gained momentum, more consumers became less sensitive to price when shopping online. Mogujie began to decline because its product positioning felt too low-end. Then Xiaohongshu emerged: a high-quality UGC community built around sharing notes and experiences related to overseas products and shopping. From there, moving into cross-border e-commerce was a natural next step. Its recommendation model often led users to forget about price altogether and shop with extraordinary enthusiasm.
That was the point when content-community e-commerce truly took off. Some platforms focused on UGC product sharing, such as Xiaohongshu and North American Deal communities. Others focused on livestreaming. Others emphasized professionally produced shopping content. Some centered on celebrity-style recommendations.
At the same time, major e-commerce platforms began copying the model by building communities and content layers of their own. Large marketplaces added social feeds, community sections, recommendation products, and livestreaming. Even the biggest platforms started redesigning around content.
The trend was impossible to miss: e-commerce platforms were trying to become content communities, and content platforms were trying to become e-commerce businesses.
Why did this happen?
Because the basic transactional needs of e-commerce had already been met by the large incumbents. Under the broader backdrop of consumption upgrading and a growing middle class in China, low price was no longer the only or even primary driver of purchase decisions. Content commerce met a different kind of need: it provided context.
In other words, user decision-making shifted from price to scenario. Content commerce sells through scenarios—showing how a product is used, what effect it creates, what kind of life it fits into, and why it matters.
In some categories, users look for recommendations about what is worth buying. In vertical fields, they look for expertise. In cross-border shopping, they look for trusted guidance. Content becomes the bridge between product and decision.
In our own case, “efficacy” or the health theme itself is a scenario. It is a scenario-based category structure. As early as 2013, I spent a great deal of time reorganizing our thousands of categories around scenarios. Unfortunately, that work was never put to use.
Product strategy: fixing the shirt before printing on it
In a book I published, I described product design as having four levels:
- Useful: identify valid needs and capture the core demand
- Usable: reshape and protect the demand, and support different use scenarios
- Easy to use: clarify structure and flow so users can operate smoothly
- Delightful to use: optimize interface and design so the product fits user preferences
Our web product, after years of refinement, has relatively few bugs. The app is a different story. It still has frequent bugs, and in terms of experience it often reaches only the level of useful and usable. It is far from easy to use, let alone delightful. That kind of experience naturally causes users to leave and not return.
And yet, while the product was still falling short at those basic levels, resources were being spread across embellishments—things like adding UnionPay support.
That kind of work is like designing a T-shirt when you still have not solved the cut and the material, but keep obsessing over the graphic printed on the front. A beautiful print may help sell the shirt, but only if the fit and fabric are already right. A plain white T-shirt is still wearable without a graphic. But a shirt with a terrible cut and bad material is not saved by a nice design.
Payment qualification issues may affect whether Alipay or WeChat Pay can be connected in certain ways, but companies the size of Alibaba and Tencent are not likely to be the reason why basic payment functionality cannot work at all.
Our product still suffers from recurring bugs: discount code issues, payment backend issues, interface issues. Some of them remain unresolved for a long time. During one app promotion, the discount code did not work, and the landing page was invalid for nearly a month. That wasted operating resources for no reason.
These decorative enhancements only make sense after the core product works. They are like a good cup of tea after a meal. If you have not eaten enough—or have no meal at all—then however fine the tea may be, it is still only water. It cannot fill you up.
The real question behind the failed KPI
We entered the China market early. We were among the first to support direct shipping, among the first to offer Chinese-language service, and among the first to integrate Alipay and WeChat Pay. So why have we still failed to make money in China?
The answer is not that the market lacks opportunity. The answer is that the assumptions behind the strategy are outdated.
China is a modern internet market. If we continue trying to attack it with the mindset of an earlier era, the job will only become harder. Eventually it becomes hard enough that leaving the market starts to look inevitable.
I am not offering a full solution here. The point is to force the discussion. If the KPI was missed, then the first thing to ask is not who failed to hit the number. The first thing to ask is whether the company ever understood what kind of market China had already become, and whether its product, data, and operating logic were built for that reality at all.