Douyin (TikTok)’s smart, personalized algorithm that delivers content uniquely for each user.
Douyin’s (TikTok’s) phenomenal success is not just because short videos make it easier to share text and visual information. More importantly, it stems from Douyin’s decentralized algorithm, which strongly motivates every content creator. This is what sets Douyin apart from earlier apps like Weishi, Bilibili, and WeChat Video Channels.
We also believe Douyin’s decentralized algorithm is still in its early stages. In the future, more industries may adopt decentralization to incentivize creators, and the level of “decentralization” will likely become even more advanced. For this reason, I want to write an article introducing Douyin’s decentralized algorithm.
Between 2012 and 2014, ByteDance’s Toutiao was already a highly successful product. At that time, it used an algorithm with logic similar to what Douyin would later adopt.
Toutiao implemented its “decentralized” algorithm by evaluating an article’s quality and recommendation value based on user completion rates:
Completion rate = Number of users who read the article in full ÷ Number of users who viewed the article
(This is similar to a video completion rate.)
The success of Toutiao provided a clear reference and foundation for developing Douyin’s decentralized algorithm.
Background
ByteDance’s first product: Jinri Toutiao (Today’s Headlines)
Jinri Toutiao → Douyin’s decentralized algorithm
Article completion rate → Video completion rate
Ecological principle → Decentralization
Nature’s evolution has no fixed center → Decentralized
“Decentralization” is the opposite of “centralization.”
In a centralized algorithm, one platform/organization decides what you see.
In a decentralized algorithm, users’ choices decide what gets recommended like a market economy instead of a planned economy.
Decentralization Analogy
Centralized Algorithm → Planned Economy
Decentralized Algorithm → Market Economy
Algorithm – Centralized – Decentralized
Economics – Planned economy (prices set centrally) – Market economy (prices set by supply and demand)
What problem does the “decentralized” algorithm aim to solve?
The “decentralized” algorithm mainly solves traffic monopoly.Instead of the platform deciding everything, a fair algorithm lets all creators get a chance to be seen, and lets users “vote” with their actions. This keeps the loop between users, creators, and platform healthy and active.
Comparing Douyin to traditional self-media platforms, we find that it’s not a platform where gaining followers early on guarantees success. Instead, creators need to consistently produce high-quality content to gain traffic. Therefore, whether in the past or present, for new or experienced creators, joining Douyin is fair; it’s never too late, which encourages more creators to join.
Bilibili
Non-followers
Followers
All followers can receive the push / notification
Douyin (Tiktok)
Non-followers
Rely on the data performance of the content itself
Followers
Pushed to followers first, but not every follower will receive it; recommendations still follow the step-by-step “traffic pool” mechanism.
To achieve a “decentralized” recommendation algorithm, the essence is to involve every user in making choices, then collect and compare the data from these choices to arrive at a result. Of course, there isn’t a direct “vote” button for users to click; Douyin (TikTok) indirectly judges user choices by collecting interaction data from content.
Recommendation principle: First, provide initial traffic to the content as a data reference for the next layer of traffic recommendation, and then upgrade layer by layer.
Currently, the recommended traffic for each traffic pool on Douyin is roughly as follows. It’s important to note that there’s still a centralized part to the recommendation process; content that gets featured on trending topics requires manual review to filter out vulgar, politically sensitive, or illegal content.
Douyin Recommendation Algorithm
Initial Traffic Pool (Cold Start)
Provides the first batch of traffic to collect objective data for evaluation
≈ 200–500 impressions
→ Question:
“Is its performance data better than similar videos in this time period?”
Yes → Second-Level Traffic Pool
Evaluate based on the performance data in the second-level pool
→ Question again:
“Is its performance data better than similar videos in this time period?”
Yes → Third-Level Traffic Pool
Evaluate based on the third-level pool data
→ Question again:
“Is its performance data better than similar videos in this time period?”
Yes → Fourth-Level Traffic Pool
Evaluate based on the fourth-level pool data
No → Stop recommending
No → Stop recommending
No → Stop recommending
Review mechanism: from broad → to strict as it moves through the traffic pools.
To compare similar videos in the same period, focus first on watch time, because it matters most. Douyin rewards content that keeps users watching, so use watch time plus other basic metrics to judge performance.
Douyin (Tiktok) Recommendation Algorithm
The Future – decentralized algorithms still have a lot of room for development, mainly in terms of horizontal and vertical integration.
To compare similar videos in the same period, focus first on watch time, because it matters most. Douyin rewards content that keeps users watching, so use watch time plus other basic metrics to judge performance.
Douyin (Tiktok) Recommendation Algorithm
The Future – decentralized algorithms still have a lot of room for development, mainly in terms of horizontal and vertical integration.
In the future, music creation, data storage, finance, and other industries may all be “decentralized.” Or, in other words, areas that are monopolized and have conflicts may be “decentralized.”
Creator income is still allocated by the platform, and we can’t yet measure everyone’s real contribution to pay them perfectly fairly. In the future this may change technologies like blockchain already offer fairer, transparent reward systems, as seen with Bitcoin miners.
What’s clear is that the old era where talent had no stage is ending. With decentralized algorithms, capable people can finally be seen. Douyin (TikTok) is like an amplifier: if you’re truly good, it magnifies you. If you’re just “1,” any power is still 1; but if you’re even slightly better, like 1.1, repeated amplification will show a huge difference. And behind every “one minute on stage” still lies “ten years of practice off stage.”
Thank you for exploring our content. We aim to provide valuable insights into the Chinese market to aid your decision-making and support your business at every step.
Our blog is your resource for Chinese marketing tips and Chinese market guidance. Contact us with questions or for more service details below.
Website: www.mmgthailand.com
Tel: 06-3167-8206
Email: info@mmgthailand.com
FB: MMG Thailand
IG: mmg_thailand
About MMG Thailand:
MMG Thailand is the first and only Chinese-owned Chinese marketing company in Thailand that aims to connect Sino-Thai cultures and power partners’ success.
We:
We’re committed to providing deep insights into the rapidly changing Chinese market and leveraging our rich experience to facilitate your growth in this dynamic environment.
Chinese outbound travel discovery is increasingly driven by interests, travel scenarios, and intent not simply by destination names. At the center of this shift is Xiaohongshu (XHS), which has become an increasingly influential.
You may have heard this claim when exploring Xiaohongshu (XHS) marketing. It sounds convincing, but it’s not quite how Xiaohongshu works.
China’s luxury market is entering a new phase. As affluent Chinese consumers shift their spending from luxury goods toward travel, wellness, and meaningful experiences, jewelry brands are facing a tougher challenge: retaining valuable Chinese customers in a slower market.
Contact us
Due to high demand of our services, we offer only a 40-minute free consultancy session
Soi sukhumvit 39 Khlong Toei Nuea, Bangkok 10110
211 Soi Pridi Banomyong 11, Sukhumvit 71 Rd., Khwaeng Phra Khanong Nuea, Watthana, Bangkok 10110
Influencer (KOL/KOC) Marketing
Xiaohongshu (Redbook) Marketing