Douyin Product Roadmap

KuaiPai and MeiPai compared to Douyin and Dashou are like Dewu and Vipshop compared to Tmall and JD.com; they are destined to be small but beautiful platforms. June 9th, 6 PM, Butterfly Dance Media.

On the second floor of this detached villa office building, there is a private office for Xia Wei, who has just finished typing the last period on a lengthy product planning document.

In the past few days, he has done nothing but focus his energy on writing this proposal.

He was originally a seasoned professional who entered the new media industry in 2017, and rose through the ranks to become a partner and vice president of Wuyou Media.

He has a thorough understanding of Douyin's products, functions, and algorithms through all the iterations and upgrades.

Douyin in 2026 will be a very large software program containing many functions.

But in its early days, it was like a newborn baby, a simple and crude soft object.

Over the course of a long decade of development, it has been continuously iterated and upgraded, adding more and more new functions, and has gradually evolved into the Douyin that everyone is familiar with.

Xia Wei's Douyin, launched in April, was just as rough as A.me, which was launched at the end of September in his previous life.

In terms of functionality, it only includes the core essential modules:

Immersive single-column full-screen playback module, shooting module, editing module, publishing module, and publishing module.

Each module has a simple function and no complicated gameplay, which lowers the barrier to entry for users.

The same applies to the algorithm; it belongs to the basic rule engine.

No complex deep learning is used; only shallow label matching and popularity weighting are performed.

Content tagging using manual annotation and keywords;

Based on user likes, follows, and dwell time, a basic interest profile is created:

The main selling point is "usable and easy to use" in order to quickly accumulate initial users.

In terms of commercialization, it was completely overlooked—zero.

Douyin, incubated by Xia Wei, successfully implemented its underlying traffic model using the V1.0 version mentioned above, becoming the fifth largest in the industry.

Douyin also has many important functions that need to be iterated and launched repeatedly.

Xia Wei clearly remembered Douyin's development path and its iterative development path.

In September 2016, A.me was launched, without complex algorithms.

Three months later, in December, version 1.0 of the algorithm was launched.

Xia Wei's Douyin (TikTok) was released directly as version 1.0, skipping the first data collection stage.

In the spring of 2017, the features of duet and music collection were launched.

That summer, the algorithm was upgraded to version 2.0.

By introducing user-content collaborative filtering and building a simple dual-tower model, the accuracy of recommendations is greatly improved.

In October of the same year, the live streaming function was officially launched.

In February 2018, algorithm version 3.0 was launched.

Major algorithm iterations have been implemented, deep learning models have been deployed, users' latent interests have been captured, multi-objective ranking has been achieved, and recommendation efficiency and accuracy have been further improved.

A month later, the e-commerce shopping cart feature was put into internal testing.

Douyin is venturing into e-commerce for the first time, connecting with third-party e-commerce platforms and launching a content-driven sales channel.

In August, the video length was increased from 15 seconds to 30 seconds, and even more were added for top influencers.

In October, the StarMap platform was launched.

An official platform for influencers to accept commercial orders, representing a significant step towards commercialization.

In the same month, Douyin Shops were launched.

In February 2019, the Hot Topics List and Topic List were launched.

This is the platform's core ranking list, which aggregates trending content from across the internet, guides traffic distribution, and creates viral topics.

A month later, the music chart officially launched.

Integrate popular background music from the platform to create a music promotion hub and strengthen the positioning of music short videos.

In the fall of 2019, the product showcase permissions were fully opened up, lowering the threshold for live-streaming sales. Real-name authentication was all that was required to open an account, driving a nationwide wave of live-streaming sales.

Starting in 2020, Douyin entered a period of maturity and explosive growth:

Live-streaming e-commerce has experienced explosive growth.

Douyin Store App Launched;

The video length is now allowed to be up to 15 minutes;

Douyin Music has officially launched independently.

Algorithm 4.0 multimodal version is now online.

Xia Wei spent five days meticulously working out Douyin's product roadmap from his previous life.

Only consider the major core functions; many minor functions are impossible to recall.

We reordered all the core functionalities and reconsidered their development and release timeline within this world.

In its previous life, Douyin's development was a process of trial and error, gradually adjusting and iterating until it finally found its way.

He has a third eye; he knows the standard answer and won't just copy it verbatim.

Instead, it will optimize and adjust many functions, change the development order, and launch new functions step by step with a better solution.

For example, Douyin's music chart feature was only launched in March 2019, following the topic chart and trending chart.

Xia Wei instructed the technology department to develop a music chart a month ago, which will be launched tonight when Douyin V2.0 is released.

It was two and a half years earlier than in the previous life, and also earlier than the trending topics and hot topics lists.

For example, Xia Wei plans to release the feature to extend video length in version V3.0, a year ahead of schedule.

We need to secure key features early on, seize market gaps, and ensure that competitors always lag behind.

The timing of live streaming and e-commerce launches will be determined based on Douyin's user base and total market size.

There's also a super-large S-level project: TikTok.

After creating A.me, the character launched on TikTok in Los Angeles in May 2017.

After its launch, it remained lukewarm and barely survived.

In November of the same year, ByteDance acquired Musical.ly, an overseas music short video app founded by Chinese, for approximately $1 billion.

In the initial stage of the acquisition, the two products maintained independent operation while simultaneously exploring integration at the technology and user levels.

At that time, Ma Weiying, the deputy director of Microsoft Research, developed a very powerful video recommendation algorithm for characters, which was used on Douyin and TK, corresponding to algorithm version 3.0.

By August 2018, Musical.Iy was shut down, and all of its more than 100 million users, content, and fan data were fully merged and migrated to TikTok.

After the merger, the "TikTok" brand name will be used uniformly, completing the integration of overseas short video businesses.

TikTok used this to quickly seize the mainstream markets in Europe and the United States, establishing its position as the world's leading short video platform.

Since Xia Wei copied Douyin, he certainly didn't want to let TK off the hook either.

There's still a full year until TK launches, so we can do a lot of things in advance.

Regarding characters, if A.me fails to launch as scheduled, or performs poorly after launch, it's highly unlikely that any TK (Team Killer) will be generated.

The thought.

Xia Wei flapped his butterfly wings, creating a series of butterfly effects, and everything changed, leading to a story completely different from his previous life.

This idea remained buried in his mind, and he didn't tell anyone. He planned to wait until he had the Series A funding before discussing it.

This involves functionality, as well as algorithm iteration.

Douyin is currently at version 1.0, having skipped the stage when A.me was launched.

Going forward, we need to skip the long trial and error process, and implement deep learning and multimodal algorithms ahead of time. We need to achieve a recommendation experience that crushes our competitors and a significant lead in user stickiness.

The key figure here is Ma Weiying from Microsoft.

Regarding the business pace, let's not rush this part; let's first solidify our core user base...

Tap the screen to use advanced tools Tip: You can use left and right keyboard keys to browse between chapters.

You'll Also Like