Obligatory Bytedance Article
The creation and future of an algorithm-first company, a conversation with Rui Ma
Welcome to the new subscribers! I write longform analysis about tech in China from the perspective of a 🐢 VC, sent out every Wednesday. If you like the content, please share with friends!
A lot of attention has been cast on TikTok and its parent company Bytedance in recent months. I originally wanted to start this piece by summarising the timeline of the events. But it looks broadly like this:
Bytedance stands out in the Chinese tech landscape for me not because it generated the fastest growth of any social media app to reach 1bn users (in the case of TikTok), or because it spent billions to inorganically acquire these users. Even though it did both of these things. I find it revolutionary since it broke away from the Copy-to-China mode of genesis for Chinese companies. Neil Shen of Sequoia (the VC who has topped the Midas List for 3 consecutive years) famously passed on their series B since he had never seen an US equivalent of Bytedance (also the market for online media was quite established). They’re that rare example of an algorithm-first company who created products around recommendation algorithms rather than the other way round.
To talk more on this I’m delighted to interview Rui Ma of Tech Buzz China, veteran tech commentator and author of an up-coming book on Bytedance. In the interview we talked about how Bytedance is AI-friendly design, their future global ambitions, philosophical differences between Chinese and Western tech companies and Kevin Mayer's regrets (or lack of).
Interview has been edited for clarity and brevity, italics are my notes added to give better context.
How would you summerise Bytedance?
I think from an organisational perspective, Zhang Yiming has managed to create a company combining the best of the West and East. This has driven efficiencies for the company though debatable whether this is as friendly for the workers. He’s created a relatively meritocratic and flat structure, which is rare in China (even in tech companies) since the structure is usually top-down and hierarchical. (Excellent speech by Zhang Yimeng on the culture of Bytedance being context over control)
The structure is great for AI algorithms which is their core engine. The product lines they generate are all shells for the algorithms (see Technode’s overview of their product suite below). Bytedance’s genesis is based on the idea that the next evolution for search is not going to be based on intentions. Search leads me to the information I want, but when there’s so much information. Is there a way to reach information of value? Making the information find you is the aim of the company currently.
From this perspective, there’s no specific category that they want to go after. Bytedance’s approach is to make a ton of apps powered by algorithms with the same back-end. Pre-algorithms, companies were built-up with a product first mentality. Post-algorithms it has been about solving questions. Questions like ‘how can we invert search?’. This style can suit Chinese entrepreneurs more since they often focus less on commercial use cases and think more about macro trends and opportunities.
What’s your take on what’s been happening with the whole TikTok saga in the US?
I’ve been really disappointed by the whole deal process. There are real legitimate concerns on certain issues like data integrity and user protection; however, they were not adequately addressed. It has been discussions on the country of origin and not about finding a solution. If you’re worried about content suppression, you should be able to audit TikTok since the info is public and you can do it without violating the privacy of the users. These and other issues such as ‘protecting the American citizens’ were not addressed. Another concern (which I raised during a testimony for the UK parliament’s online harm committee) was the lack of understanding on the long term impact of machine curation and recommendation on the population.
The media was also often unhelpful; they made it about political stances. It felt irresponsible when some of them were interviewing international relationship experts rather than security experts for insights on this issue. No one seemed to care about the actual issues. Even people in Silicon Valley were uncritical of the media narrative and clearly for a lot of people the main reason became reciprocity. This was disappointing, because in my opinion this helps legitimize bad behavior in the future. Furthermore, the views were also very uninformed and came from people who have never done business in China.
What did you think of Bytedance’s responses?
It made no sense for Bytedance to sell, I think they should shut down US operations instead if it came to that. Since AI is the core offering and TikTok is just a shell for all the learnings, shutting down TikTok would allow them to protect the algorithm. Not selling TikTok also protects Douyin from potential contenders from buyers who might launch a Chinese offering after buying TikTok.
I understand why the Bytedance board was considering the offering since that is good fiduciary duty to explore all the options. But I think it’s unlikely to have a majority sale including the algorithm at this point.
What’s your impression of what Bytedance will do from now on? Do you think they will diverge from their vision of being a global company?
Zhang Yiming has learnt his lesson and will approach future incidents from an understanding that ‘this is a problem I can solve’. This is a battle that they can win once certain conditions are met.
I think he’s still going to after the rest of the world. If he doesn’t, his competitors will, and they will get bigger. A bigger company will mean more access to capital and more data, which are significant moats for any tech company. Being a global company also means that they can service their Chinese customers better, since most big brands in China are international brands. Also, when you’re using global talent to service a global population, you’ll be a better company.
In the short term, they’ll adjust their strategy for the US, India and Australia but there are other places for them to grow. Europe just hit 100m MAU for example. TikTok’s success has only proven it can be done.
What do you think Bytedance’s core capabilities are?
Bytedance focused on the recommendation off the bat and they have commercialised that well through different apps. Most people (especially in the West), were working on personalisation as a way to curate and not recommendation. Bytedance was able to identify the right obstacles to make recommendations work and didn’t tie it to a specific product, but rather make it applicable to all entertainment (and beyond). TikTok was their focus on short-form entertainment and they were very successful with that iteration.
The way that other startups approached the topic, like Flipboard back in the days, was using personalisation based on your social networks. However, the personalisation experience didn’t get better the more you used it. Bytedance’s algorithms get better with every single use.
Bytedance was sued for piracy in the beginning (for Toutiao) as their contents were ripped from other websites and displayed with different advertising. Either by choice or forced to, Bytedance learnt to cultivate and invest in a creator community quite early on to sustain themselves, which is something other companies didn’t understand. Youtube grew organically while Bytedance grew Douyin and TikTok intentionally (with both advertising and also invested in the creator community early on).
It’s tough to succeed with doing a lot of different things in different segments in the West. I’ve noticed that it’s relatively common in China with all of the BAT (Baidu, Alibaba and Tencent) and TMD (Toutiao/ Bytedance, Meituan and Didi) doing variants of multi-sector product offering - are there particular reasons why this conglomerate style thrives in China?
I think the number one factor is that Chinese companies think about the user experience, and not so much what the app is delivering. Companies think about what users are doing on their phones and what else they can do to get more of users’ time. That’s what Tencent has done with games after WeChat (see my post about Bilibili for another example).
The second factor is that Chinese companies have an ecosystem mindset. While western companies think about products, Chinese companies think about the chain that makes the product a success. Take livestreaming e-commerce, to make that business model successful, you need to have a lot of professional livestreamers. To get a lot of professional livestreamers, you need training schools for livestreamers. Bytedance has invested in MNCs (influencer and livestreamer agencies). Bilibili has verified MCNs that they work with. Chinese companies boost the ecosystem in this way rather than relying on it to appear.
Meituan’s progress reflects these two factors well. They deliver food but have expanded into hotels and flights. Since the user will be likely to book hotels and flights with them if they are already customers. They have also integrated their order management and delivery software with restaurants’ POS to generate a more seamless experience. So they do both vertical and horizontal integration. In the US, companies will typically pick one form of integration or the other.
Another difference is that western companies will go for the highest ROI action, in contrast Chinese companies may pursue activities which might be loss-making in themselves but have positive synergy when combined (such as Alibaba’s extensive logistic networks and payment ecosystem are to facilitate a frictionless ecommerce experience). Chinese companies tend to look at things more holistically and leave more room for second-order effects more in their decision making. Of course, they’re often wrong as well.
How do you think the US government’s issues with Bytedance affects the ambitions of the Chinese internet?
It’s too early to tell, but I don’t think companies will stop their international expansion plans. Since if they don’t do it, their competitors will. Giants like Alibaba and Tencent will keep going with global expansion since their yield in China is diminishing.
The question is, will the small startups change their strategies? It will depend on the region, they probably wouldn’t want to invest in the US and India right now. The tide of globalisation wouldn’t stop, but for certain regions it will decrease.
If all these interesting Chinese apps aren’t available in the US but are available elsewhere in the world, then the US consumer is the group that’s getting hurt. TikTok is part of the next evolution of jobs and has brought new opportunities to their creators (Note: see Lil Nas X’s Old Town Road and Sarah Cooper’s new Netflix show). You now have professional Youtubers and that category of jobs will only get bigger. This is bad for the younger generations whose job opportunities are getting more restricted.
Last question, based on your great tweet:
Do you think Kevin Mayer regrets his resignation?
It depends on how much he was tied to success in America, how much he defined success as being successful in America.If he wanted to be a big deal in America, then no. If he wanted to be a global CEO for a company that’s as big as Facebook, then he would regret it. Not everyone cares about being a global presence. (Also seems like Kevin’s doing ok)