Below is the edited transcript of Boyu’s speech in XVC Annual Conference 2020:

Hello, everyone. Thank you for attending XVC annual conference 2020. It has been four years since the initial close of our first RMB fund and three years since the initial close of our first USD fund. By October 31, 2020, the total cost of our investments was USD 212.5 million and the market value was USD 393.5 million, representing a gross IRR[1]48.15%。

2020 has been a tough year for most of the countries due to COVID-19. Very few of us have experienced something like this. You may wonder how we are doing in 2020?

Today, I will answer some of the most frequently asked questions in this year, share our observations on some structural changes, and explain to you our strategy in response to the changes.

First, let’s look at the Chinese VC market in 2020.

This chart above shows VC investment activity for the first three quarters of 2020. You can see a double-digit growth in capital deployed by VC funds in the first three quarters of this year. The activity was very low during the crisis, so the rebound is a full recovery, and the number is aligned with what we felt in the market.

And now comes the question-probably the most frequently asked question in 2020. What is the impact of COVID-19?

Our answer: COVID-19 had a very big impact on economic behaviors in most countries in the world, but this impact is likely to be short term. It seems vaccinations are coming as we hoped, and they actually work. Above all, it seems that with or without vaccination, the economy is still growing.

This chart shows the combined Revenue / GMV of our portfolio companies. We can see a dip in Q1, which is mostly due to Chinese New Year and Covid-19 , then a quick recovery in Q2 and Q3.

This is a stress test we did before. We put the active portfolio companies into a 12-bucket matrix. One dimension is cash flow and profitability, and the other dimension is cash reserve. As you can see in the chart, most companies have a very long run way of over 24 months. And a few companies are profitable or cash flow positive.

Now let’s spend a few minutes on the second most frequently asked question: What is the impact of U.S.-China tension. Should we expect a slow-down?

Since the U.S. China tensions began, some people developed a theory saying that China would repeat what happened in Japan in the 1980s. This comparison was made, because the China of today resembles Japan in the 1980s in many ways, including trajectories of growth, aging of the population, and external pressures. Based on these resemblances, many people argue that China will become another Japan, facing the same difficulties in upgrading its industries and entering the so called “middle income trap.”

You saw these same charts last year. In our eyes, things are not this simple. We see at least two Chinas. The gray bars represent the average disposable income of the top 300 million people in China with above average disposable incomes, and the red bars represent the average disposable income of the remaining 1.1 billion people in China. We’ve updated the data in the two charts. In the chart on the right, you can see that 2019 was still a strong year for these the 1.1 billion people, as their disposable income grew at over 10%, and this happened during the trade war with the U.S.

Now let’s get back to our comparison with Japan. As you can see, China is unique in that it covers a massive area and has many different-sized cities. In fact, China’s major cities are so large that a single metropolitan area can be larger than a country. And it has hundreds of smaller cities. Here, we used a slightly different approach to display the data. The top 30 Chinese cities have approximately 300 million residents. Their GDP per capita is represented by the blue line, and the remaining 1.1 billion people are represented by the red line. The gap between the two represents about 10 years of growth. Japan’s gap between large cities and smaller cities is much smaller. As a result, the income of these 1.1 billion Chinese people is likely to maintain a very high speed of growth.

In fact, the urbanization rate in China is around 60%, which is 20 points below the consensus equilibrium. This part of the country’s urbanization will mostly occur in lower-tier cities. As we noted at last year’s AGM, the spending of these lower-end 1.1 billion people has been increasing much faster than the GDP. In the future, these citizens will become the main drivers of China’s economic growth. Moreover, these 1.1 billion people are not only consumers, but they are also producer. These 1.1 billion populations allow China to develop and grow its internal economy cycle and industry clusters, and industry clusters have network effects. A complete set of industry clusters needs the support of a very large population.

Therefore, we don’t think China will be another Japan. China is in a much better position to continue to upgrade its industries and grow its consumption and production. In the next decade, we believe China’s line on the growth chart will look much better than Japan’s.

However, we are entrepreneurs and venture capitalists, not asset allocators. It is our job to find next-gen USD 10 billion plus opportunities and make 50x plus return on them.

We have to find and act on some structural changes that can have a greater impact on economic behaviors than COVID-19 and U.S.-China tension can. So, next, I will explain to you some structural changes that we have observed.

The charts above show some data we have collected. The chart on the left indicates the change in the average time that people spend on instant music videos per day. In China, for instance, the music video is “Duan Shi Pin” (or short video). Most viewers’ screen time was spent on Douyin and Kwai, two music video apps. Think about it - on average, people are spending 98 minutes per day on Douyin and Kwai. Have you tried doing it? I have to admit that there were a few times, when I lied in bed, I couldn’t stop myself from flipping the screen and watching one video clip after another. I could feel my brain getting scrambled like scrambled eggs. But I couldn’t stop it. These two apps are probably the most powerful brain washing machines built in human history.

Let’s look at the chart on the right. The advertising revenue of ByteDance and Kwai is expected to be 2x of the combined ad revenue of all TV stations (or TV networks) in 2020, if numbers are accurate.

Now let’s get a sense of the magnitude of the structural change in China’s retail channels. These are some new retail channels that barely existed just four years ago. From bottom to top, these bars represent the GMV of Taobao live streaming, Kwai, Douyin, WeChat Mini Program, and PDD. The total GMV of these new channels is expected to be about USD 614 billion. As a benchmark, the total GMV of Amazon Marketplace was only USD 200 billion in 2019. That’s the total GMV of Amazon Marketplace, and that number is only 70% of the incremental GMV of these new channels in 2020 alone. Let me put it another way. They grew a new Amazon Marketplace in less than three quarters. And if you look at the small grey part, we believe it will likely become much larger in two years. That’s Douyin’s market place, and it has just started.

WeChat is becoming the largest CRM system that brands are using to serve and retain customers. The numbers in the chart on the right are taken from the prospectus of Perfect Diary. The percentage of newly acquired customers came back for a second purchase within three quarters almost quadrupled in 2018, and we believe much of this growth should be attributed to the use of WeChat as their CRM platform to serve and retain customers. This is called “private traffic” (in Chinese “私域流量”).

Let’s do a case study of Perfect Diary. Although I think it’s going to take a little more time to draw the conclusion that Perfect Diary is a great company, I believe it’s fair to say that it’s extremely innovative. To better understand what Perfect Diary did correctly, let’s first review how P&G built and strengthened its market leadership in its time. This is found on the left chart.

P&G’s first step is always finding the “common denominators.” In other words, P&G finds products that satisfy the needs of a big enough crowd; otherwise, the products won’t be able to get on the limited shelf space of retail outlets, most of which are modern trades, such as super-markets.

The second step is to leverage their sales team and relationships with retailers to get the products on as many shelves as possible.

The third step is to initiate a saturated attack on mass media, like CCTV, to build brand recognition and, thus, increase product turnover at retail outlets.

However, this model has three disadvantages:

1.      The products it sells must be targeted to very large user groups, making it difficult to precisely target many small user groups with different needs.

2.      The brand’s knowledge of users is very limited, and the iteration cycle of its products is very long and slow.

3.      It lacks channels to directly interact and retarget its existing customers, let alone reach them competently.

The old way is more like farming, and we see a new way which is more like fishing.

The new fishing mechanism is based on 1) direct interaction with customers; and 2) behavioral data and algorithm-driven strategies.

This is how Perfect Diary acquires and serves its customers:

First, it finds unmet needs through behavioral data and quickly launches beta products.

Second, it tracks the testing data to find the optimal “infection path”, target-user portraits, and influencing content. Sometimes they even use algorithms to find the right combination of KOLs to create a marketing “inflection point”.

Third, they scale the marketing using “look-alikes” to retain customers in the “fish pond” (a private traffic pool, usually in WeChat) and tag broadcasters and customers using their behaviors.

Finally, they use behavioral data and self-adaptive algorithm to improve the efficiency of targeting and re-targeting to drive better acquisition and retention. As they keep doing these things, they will continue to get better at it over time.

In this way, the brand learns something about each customer from day 1, and then they learn a bit more each time the customer interacts with it. In essence, each customer is training the algorithm to know how to better market or serve other customers who look like them.

Also, using cross-platform retargeting technology, the brand can reach customers in various formats in different places (i.e. WeChat moments, mini program, streaming, videos). Marketers can increase brand awareness, interest, and purchases in a multidimensional way.

Let me show you another case. This is one of our portfolio companies. They use data and algorithms to identify unmet needs, test ideas, and optimize the supply chain and inventory with nearly instant data feeds from the marketing department. Then, they use data and algorithms to brainwash consumers;  for example, they single out hundreds of KOLs and KOCs out of tens of thousands and created various brainwashing content for each of them with the help of data and algorithms. They have also developed the skills to create video clips that can pass the test of Douyin’s algorithm to get a large volume of views, and then used WeChat to be their CRM portal, using robots to help customer representatives serve customers.

We are seeing a new world with new rules. Consumers are changing, and the channels to reach, serve, and retain them are changing even more quickly-especially the ways to establish and strengthen brand associations. The new world is ruled by data and algorithms, but the current supply is old and slow. They are not only managed by old people; they are still running with the same old rules in the same old organizations that used to make them successful. But today, the rules are changing so rapidly that we believe new companies created by people who understand these new rules will be the next-gen rulers of the new world.

Now let’s take a look at our investment theses. This is a very old slide that I’m sure you have seen before. It stayed almost the same since we started raising our first fund four-and-a-half years ago. If you are not tired of it yet, you will be, because you will probably see this slide again at every annual meeting in the next 10 to 15 years. After that, you will probably see something different, because at that time, this presentation might be delivered by someone else, someone younger and smarter than me.

You might wonder: Didn’t you just tell us that it’s a new world? How come your investment theses haven’t changed? Basically, this slide illustrates our understanding of how a business can be friends of time in a market economy. What a good business should look like has not changed during the last 100 years, and I believe it’s not going to change in the foreseeable future.

At XVC, we focus on four types of opportunities: platforms that have network effects; applications that can continuously generate or monopolize data and use it to improve their own user experience; products or services with economy of scale; and self-reinforcing brands. This year, you can see that we made a number of investments in the fourth category. We have talked about the first three categories a lot in our previous annual meetings. Today, I’d like to spend some time on the fourth category: self-reinforcing brands.

So, what exactly do we mean by “self-reinforcing brands”?

Let me start explaining this concept by showing you some research. This is research done by two scholars in Germany. It’s simple. They offered four cups of identical drinks to test participants (the four cups contained an equal mix of three cola products: Coca-Cola, Pepsi, and River Cola). They told participants that the drinks were one of  four brands: Coca-Cola, Pepsi, River Cola, or T Cola. “River Cola” is a generic brand sold by a big German supermarket chain, and “T Cola” was an invented brand that doesn’t exist. In this case, Coca-Cola and Pepsi represent strong brands, while River Cola and T Cola represent weak brands.

So, this is how they did it. Each test participantrepeated the following steps 48 times: They looked at the brand cue, put 1 ml of the drink into their mouth with a tube, kept it in their mouth for six seconds, swallowed, and waited for another six seconds. Then they rated their satisfaction (levels 1 to 8). As they completed these steps, the test administrator continuously scanned their brain activity with fMRI.

The results are quite interesting. First, none of the test participants noticed that the drinks were identical. What’s really surprising is that the participants not only reported higher satisfaction after they drank the drinks that they thought they were drinking, but their feelings of increased satisfaction were actually signaled by higher neural responses in the area signaling reward in the brain, as shown in the brain scan. In other words, the strong brands themselves created incremental satisfaction.

The chart on the left shows the number of retail outlets penetrated by different companies. By 2011, Coca-Cola was serving 140 million consumers every day through 8.8 million points-of-sale. If Coca-Cola were willing to place some ads on its bottles, it could probably make more money on advertising than most newspapers and magazines in China. I believe this “brand loyalty on a massive scale” is the main reason Coca-Cola  can survive in so many sales channels, including the smallest restaurants, mom-and-pop shops, or vending machines. And these 8.8 million retail outlets are helping Coca-Cola acquire new customers and strengthen its brand loyalty every day. Just think about it: 140 million customers served per today, and this is only in China (it has 30 million retail outlets globally).

As you can see, Coca-Cola is a good example of one of the several types of self-reinforcing brands.

We classify self-reinforcing brands into three categories: consensus locking brands, happiness imprinting brands, and trust and safety imprinting brands. Again, let me give you some examples.

TAL Education Group started as a tutoring school for kids to compete in math Olympics contests. In the cities they are in, they can always make sure the smartest kids would enroll, so their students could usually win most of the medals in the math Olympics. This is how they created a self-reinforcing brand. Let me ask those of you who are parents: Would you send your child to a tutoring school that produces math Olympics gold medal winners but requires passing a very hard test to enroll? Or would you send your child to a different tutoring school that’s much cheaper, but the students are not as smart or don’t work as hard to win medals? Well, I think my kids are smart, so I would send them to TAL Education. As you can see, this company successfully created a consensus: the smartest kids go to TAL Education. Others go to less competitive tutoring schools.

Another example is Moutai, which is the most expensive Chinese baijiu, a special distilled Chinese liquor, that Chinese people serve to their guests. The consensus is that: you should treat your most distinguished guests with Moutai and treat other guests with other baijiu.

Consensus locking brands are rare. Most brands fall into the second or third category: happiness imprinting and trust and safety imprinting.

Happiness imprinting brands can create incremental satisfaction like Coca-Cola does, as we just discussed. Some brands achieve this by offering unique taste, plus, of course, a lot of brainwashing advertising and massive shelf displays. Other brands do this with scale and a density-enabled supply chain. For example, ZARA, an apparel retailer can offer extra satisfaction through a unique shopping experience. You can always find lots of new arrivals, trendy and affordable clothes in their stores, so you would stay in their store and buy a lot, while ignoring the other smaller brands nearby. This brand is powered by its strong supply chain, which relies on scale.

Another case is Domino’s Pizza. Nobody likes cold pizza, so, the fastest and cheapest way to deliver pizzas to customers is to hire dedicated delivery people. Each delivery person would carry multiple orders every time he or she leave the store. A smaller pizza brand can’t do this because they don’t have the customer density. The delivery person would either take fewer pizzas per route (which could take longer to drive farther-meaning the pizza orders gets colder) or rely on a third-party delivery system. Imagine how long it would take if a third-party delivery person had to pick up orders at multiple places and then deliver them all.

The third type is trust and safety imprinting brands. We need these brands because, frankly, shit happens. Our IT systems are poorly designed, full of bugs, and sometimes hacked. Our kids pee and poop in our cars. We sometimes get sick, and some people die of sicknesses. What a dangerous world we live in! So we need trustworthy products to solve our problems. Humans are risk-averse, especially to matters that can have bad consequences. So, a leading brand can leverage higher channel penetration, better shelf display, and a larger brainwashing budget to strengthen their customers’ brand awareness and trust, which in turn strengthens the brand and crowds out competitors.

Self-reinforcing brands can leverage their brand leadership to build more efficient channels to acquire, serve, and retain customers, which strengthens their leadership. As a result, a less known brand would have to stay away from the leading brand and differentiate itself by choosing a different segment, usually smaller and less lucrative. Here, “channel” means all the intermediaries needed to acquire, serve, and retain your customers, including media, distributors, retailers, franchisees, or landlords (if you open your own stores like ZARA, Starbucks, and Domino’s), as well as your own sales team and customer service team. For some consensus locking brands, their existing customers are their largest channel to acquire new customers. For example, our portfolio company, Zhuangxiaomi, gets 90% of its customers by word-of-mouth, and Hetao coding gets 60% of new customers from word of mouth.

Let’s study two cases in our portfolio: one is a leading cosmetic contact lens brand (Company M). You know what people do with contact lenses, right? They put these products in their eyes. Well, how often do you put something in your eyes? You need to trust that this product is not going to blind you. Therefore, Company M started with high-end daily disposable products to create repeat purchasing and positive word-of-mouth. But here’s something interesting with this category: About 70% of people in China have myopia, like me. They also have different degrees of myopia. Colored contact lenses need to be adjusted for that. That’s why they need a large number of SKUs, due to the multiple colors and degrees. Also, people need new stuff, new colors, new designs, a fast-fashion type of consumer behavior. Consequently, these products created a very strong economy of scale on the supply side. A bigger player can manufacture products cheaper and faster, and it's much easier for them to keep inventory. Also, KOLs are risk-averse in this industry. They are afraid of getting into trouble, so they only work with leading brands. That’s how Company M will reinforce their brand going forward.

Another case is a fast growing baijiu (Chinese liquor) brand (Company G). As many of you probably know, lots of baijiu consumers are loyal to a flavor. They can be loyal to one flavor and one product for like 30 years. Company G identified the unmet needs of next-gen baijiu consumers and created a product and packaging to meet these needs. Guangliang has been very smart; starting with storeowners, the company spread social consensus and gradually increased product turnover by gaining repeat customers. Then they leveraged their high turnover to lock in retailers’ shelf space and distributors’ capacity to crowd out competitors. Their channel efficiency and their brand loyalty reinforce each other.

These self-reinforcing brands bring us a new challenge, because without the network-effect type of first-mover advantages, it takes longer to build a “moat”. Our answer to this is three-fold.

First, we will build an information network to continuously observe and understand the market, being a little more stage-agnostic. This year, you can see that we have made a number of investments in the seed stage and also a number of investments in the growth stage.

Second, we’re going to partner with great organization builders, not with people who only have ideas.

Third, we will develop skills to help the founders develop organizational skills.

After all, great businesses are all great organizations. Some great businesses were good businesses from day one, but most of them were not. In fact, if you check out these logos-Huawei, Alibaba, Pinduoduo, Meituan, and ByteDance-all of these companies are now doing almost completely different things from what they were doing when they started their businesses and came out to raise their series As and Bs. However, these are great organizations created by great organization builders, so they are able to reinvent themselves and catch new and bigger opportunities whenever they see one.

And these companies, they didn’t inherit or find a moat. They built their own moat. But they didn’t stop there. They kept exploring their boundaries. They know how to attract and retain the best talent in the market, and continually remove unfit and underperforming employees from their organizations. They know how to build effective incentive systems and good cultures to ensure that employees are motivated and hardworking, and that employees at every level can make good decisions aligned with the long-term success of the organizations.

To understand how entrepreneurs build great organizations and to help our CEOs develop these skills, we need to learn how to do it ourselves.

This slide above lists some things we did and some things we will try. We have hired experts to teach ourselves how to conduct proper behavior-based interviews and asked them to shadow our interviews. We now shadow interviews for each other and also for our CEOs. We practice defining and refining “competency models” and “talent portraits” of some positions with headhunters and CEOs. We also use a mix of SOP, KPI and OKR and periodical reviewing to build a self-improving feedback system within XVC’s mid-back office. We have been continuing to hire good people, while unfit and underperforming employees have left the firm. This has created a self-improving cycle within the firm. We have defined XVC’s mission and company culture with explicit words and embedded the values into our processes to turn them into consensus and habits.

We’re going to try something new, too. We will first extend the OKR system to our deal team. Then, we will upgrade XVC’s mid-office, building three teams and extending their services to pre-termsheet deals and our portfolio companies. These three departments will be named: Research & Strategy Team, Finance & Risk Team, and HR & Organization Team.

Does this make sense to you? Does this sound like what venture capital firms should be doing? Are we distracted and not focused enough? Well, we ask ourselves the same questions. And then we asked another question, a bigger one: Who are we? What is XVC?

And this is our answer:

We are independent thinkers. We are fact-driven researchers and decision-makers. We search for self-reinforcing business models that can be friends of time. But we don’t want to just free ride the entrepreneurs. We’d like to try our best to help our companies win and help our CEOs grow. We are comfortable taking risks. We understand it’s important to stay focused, but we will not stop exploring our boundaries and trying every means possible to create long-term value, and that includes becoming good organization builders ourselves. This is what we mean when we say, “We are entrepreneurs who happen to be venture capitalists.”

Finally, I want to thank you again for supporting XVC. We feel blessed to have this group of long-term thinking investors so we can focus on doing our job. We look forward to a long-term and rewarding partnership with you. (End)


  [1] The Internal Rate of Return (IRR) calculation uses remittance dates for cash outflows, and it assumes we liquidated the remaining portfolio on 10/31/2020 at its latest market value. We wrote off some companies in hibernation mode, and the final number has factored in the write-offs. The investment cost and market value included around $20 million USD of liquid assets in the brokerage accounts of the USD funds.  


Appendix: