ByteDance Is Going for Fast, Decentralized Innovation at Global Scale (Tech Strategy – Daily Article)

I recently visited the ByteDance office in Beijing. And I left with 4 take-aways about things I had not really appreciated about the comapny. Those are located here.

My biggest take-away was that ByteDance is facing a really interesting organizational question. The company is both more innovative and more decentralized than I had realized. And it was increasingly operating this way as a multinational at global scale. That’s an interesting question.

Can a company do rapid decentralized innovation once it becomes very large, very multinational, and very successful?

Can it continue to be highly decentralized and innovative at this phase of its existence?

My thinking on this is 3 points (so far).

Point 1: Entertainment-Based Attention Is the Fastest Changing Sector in Digital

ByteDance has aptly been called the “attention factory”. And it is doing something that is inherently difficult. It is in the business of getting people’s attention. And keeping it.

That’s really valuable if you can achieve it. But it’s also really hard.

Because we now live in the age of abundance, where there are endless products and content to be consumed. Supply keeps increasing while demand is largely fixed. There are millions of videos on YouTube but only 24 hours in a day. Attention is becoming the key scarce resource. And it is increasingly valuable.

Companies that had achieved the sustained capture of attention are usually offering a utility service. Payment services get frequent engagement and attention. So do social media and communication services. Pretty much everything Google offers (search, email, maps) are frequently used utilities. Utility services you require every day are a good way to get attention.

But ByteDance is doing something much harder that this. They are capturing sustained attention with entertainment and media. They are keeping people engaged by continually giving them things they like to watch. Not something need. Something they want. They are good at showing people the content they most want to watch in a given moment. It is personalized. And it keeps changing.

Capturing attention with entertainment is a dynamic, never-ending process. What a person wants to watch today is different than yesterday. In fact, it is usually different than an hour ago. And it is keeps changing at the individual and demographic level, with new trends and habits always rising and falling in media and entertainment.

Capturing attention with entertainment is also dynamic in terms of media formats. People really like short videos right now. But they didn’t ten years ago. What consumers might want to watch in five years is unknown. Maybe AR/VR? Maybe online conversations and podcasts?

My first take-away is that entertainment-based attention is the most rapidly changing sector in digital. You have to be really agile and innovative. You have to keep adapting to never-ending changes in terrain and behavior.

So how has ByteDance achieved this?

In my opinion, ByteDance achieved this by doing two important things:

First, they has created a new mechanism for the interaction between human beings and the sea of content on the internet.

There is a gap between our brains and all the information on the internet. How do you find what you want to see or read? How do you choose what to show on the smartphone screen? Which is the very limited connection between our brains and the internet?

Google Search provided one good solution to this problem. Search is smart and scalable. It was a big improvement on the portal websites (like Yahoo) that came before it. But it requires active searching by the user. And humans get only a smart percentage of their information from active search.

Facebook provided another solution to this problem (sort of). It let users scroll a newsfeed, which teed up relevant information and content. This is also a scalable and smart solution. And it is passive consumption, as opposed to active searching. But it was based on your social network, which is a crude hack to determine what content you might want to see. A newsfeed based on a social network excludes the vast majority of long-tail content (which your friends aren’t posting).

Enter ByteDance, which pioneered AI-based matching between you and content on the internet. It’s a newsfeed which means passive consumption. You don’t have to actively search. And it doesn’t care who your friends are. In fact, it is so effectively, it can show you what you most want to see with no input beyond the flicking of your thumb. It was a bold solution to a major tech problem. And it worked with news headlines (i.e., Toutiao). And it is working well with short video, which is why TikTok is so addictive. If ByteDance is an attention factory, this AI matching is the “attention” part.

Second, ByteDance created an organization that is constantly creating and testing new mobile apps.

Media formats, technologies and behaviors keep changing. They are almost impossible to predict consistently. So, you want to do lots of experiments and adapt quickly. You want to create and test tons of mobile apps all the time. This is the “factory” part of attention factory.

This is how ByteDance discovered news headlines (i.e., Toutiao). It is how they discovered short videos (i.e., TikTok). ByteDance is constantly creating and testing hundreds of apps to stay on the frontier of changing human attention.

Much has been written about ByteDance’s matching algorithms. But it’s mobile app factory that I think is the most interesting. This is about the company’s organizational structure, human capital, and culture. This is actually much harder to understand and replicate than the algorithms.

Which brings me to point 2.

Point 2: ByteDance Is Doing a Faster Version of the Decentralized and Agile Approach We See at Alibaba and Amazon

Creating an innovative and agile organization is a big topic in digital transformation. In practice, it usually means moving operational and innovation activities to data-driven and autonomous teams. This type of team-based organizational structure is usually more natural for companies “born digital” than for traditional companies that are undergoing a difficult transformation to digital.

Probably the most famous example of a data-driven, modular, and team-based organization structure is Amazon. They were an early adopted and some well-known books have been written about their notoriously flat (3 level) organization structure. They have hundreds and thousands of teams (sometimes called pods) working on various products and foundational technologies. These teams are run by product owners and typically contain 6-12 team members with a mix of business, data, and technology skills. These teams are how the company continually improves customer-facing products and features. That’s where the rubber hits the road. You push as much authority to teams that are close to the customers. They are the front lines in a fight to make continuous improvements to the customer experience. This can be everything from major upgrades to removal of points of friction and inconvenience.

This team-based approach is also how new products can be created, especially mobile apps. The common approach is to empower product teams to rapidly create new apps and then test them with limited launches. Those apps that get traction can then fueled and grown. Those that don’t are culled. And those few apps that reach significant scale can then be turned into mature businesses.

This is sometimes called an “explore and exploit” approach to product development. Alibaba depicts this product development process as a series of bushes and trees.

From their investor day presentation:

Note how they separate their products into seeds, traction, and profitability. That’s pretty cool.


So, what happens when you apply this decentralized, agile approach to the rapidly changing world of entertainment-based attention?

Ecommerce develops rapidly. But not nearly as quickly as entertainment. How people buy staples like socks, detergent, and computers doesn’t change that much. It is not nearly as dynamic a space as short videos, podcasts, music, and other forms of entertainment.

So, what ByteDance has been doing is a much faster version of the organizational approach of the ecommerce giants like Amazon and Alibaba. And ByteDance is has been very successful at this. Look at their growing list of very successful mobile apps:

  • Toutiao
  • Xigua
  • Douyin / TikTok
  • Lark
  • CapCut

And now we see Volcano Engine and Lemon8 emerging.

Which brings me back to my main question.

Point 3: How Will ByteDance’s Approach to Decentralized Innovation Change Now that the Company Is a Large, Successful Multinational?

Much of ByteDance’s speed and agility in the past came from being a much smaller organization, with deeply involved founders. If scale is the biggest advantage of dominant organizations, speed is usually the biggest advantage of smaller ones.

But now ByteDance has +115,000 employees and is spread across in 30 countries. They now have lots of advantages from their large scale. But there are also disadvantages to scale. Larger organizations are prone to bureaucracy. Larger organizations struggle with the increasing complexity of the product mix and operations. They are often vulnerable to more specialized companies.

What’s more, ByteDance is also now a multinational. TikTok in the US is building its own products. Lark is building in SE Asia. And so on. Being a multinational is harder organizationally than being a single country company. There are more cultural, political, and organizational complexities.

And finally, there is the problem of being really successful. Smaller companies tend to be more focused. Management and operations tend to be leaner and meaner. They have to be. They have limited resources and are focused on survival.

As companies become large and successful, there is a tendency to become organizationally bloated. It’s too easy to just greenlight lots of projects. It’s easier to lose focus. And it’s really easy to lose the lean and hungry culture of a smaller struggling company. People stop coming into the office at 9pm on Saturday nights. Silicon Valley is full of very large and successful companies that have become slow-moving bureaucracies. Facebook hasn’t created a successful new product in about 15 years. Neither has Apple.

In contrast, Huawei is actually a great example of a company that actively fights against the complacency of large size and big success. It always looks for competitors that can challenge it (historically this was Ericsson). It actively creates compelling and challenging career paths for its younger staff. It even forces senior management to exit, creating open senior positions for younger staff to aspire to. As I am writing this, Alibaba is reorganizing into six separate business units aligned with capital markets. This is an attempt to re-energize the operating performance of a large successful business.

I don’t see any indications of slowing down at ByteDance right now. But it is part of a really interesting question. How can ByteDance continue to rapidly innovate as a large, decentralized, and successful multinational organization?

I’ve been trying to get better visibility into how their newer products have been created. Services such as CapCut, Lark and Volcano Engine. As an outsider, I have a couple of working observations:

  • They like to build new mobile apps and services from their internal capabilities. Lark and Volcano Engine were internal capabilities that were then offered to the market. It’s also interesting that they are focusing on enterprise customers as well as consumers now. That’s a different type of attention.
  • They like build on top of existing mobile apps. That’s CapCut, which builds on TikTok and is a fantastic tool for content creators. The filters alone are worth downloading it. It’s also interesting that they focused on content creators instead of consumers.
  • Products are global but tend to originate in China. The China market moves very fast so it’s natural for them to emerge there. But most of these are global products that are ultimately = managed by teams in local markets. I’m wondering if we are going to see more new products originate in the US and other places.

That’s where I am in my thinking right now. I’m watching ByteDance closely. Human resources must be really interesting right now.

Cheers, Jeff


Related articles:

From the Concept Library, concepts for this article are:

  • n/a

From the Company Library, companies for this article are:

  • ByteDance

Photo by Alexander Shatov on Unsplash


I write, speak and consult about how to win (and not lose) in digital strategy and transformation.

I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.

My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.

This content (articles, podcasts, website info) is not investment, legal or tax advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. This is not investment advice. Investing is risky. Do your own research.

Leave a Reply