The product manager of NetEase Cloud Music personally tells: practical experience in each product development stage of Cloud Music!

The topic I want to share with you today is "The practical experience of Cloud Music in various product development stages"

Why do we want to do NetEase Cloud Music?

Last year, the number of NetEase Music users exceeded 100 million. Why do you want to do NetEase Cloud Music? Ding Lei, the founder of NetEase, is a very senior music enthusiast. He did not like most of the music apps on the market at that time because resource sharing was relatively scarce. So he wanted to make a great music app in the era of mobile Internet , which was the starting point for us to make music.

From my perspective, when I was making this product, I would think about how this would help the entire music industry, how it would help NetEase, and what kind of convenience it would bring to users.

Kugou, QQ Music, Xiami Music

Faced with numerous strong competitors, how can Cloud Music break through?

So what issues did our team consider before creating NetEase Cloud Music?

1. Conduct market analysis to see the strengths and weaknesses of your competitors

When NetEase Music was first launched three years ago, there were quite a lot of music apps on China's mobile Internet , the top ones being Kugou, QQ Music and Kuwo Music. The second camp basically includes Baidu Music, Tiantian Dongting, Xiami Music, Douban FM, etc.

When analyzing the market, we found that it is a large market that lacks innovation.

Why do I say so? Although the above products have been in the industry for a long time, it took five or even ten years to accumulate 100 million, 200 million, or even 300 million users. But if you look closely at these products, you will find that they are all music library products, and their main functions are searching and updating some songs.

Users mainly play music by searching and editing hot keys. The first conclusion we have reached is that music products on the Internet have been developing for such a long time, but there has not been any particularly major innovation.

2. Surveying users, the conclusion is that users are becoming lazier

In the PC era, users were more accustomed to products like music libraries. The biggest actions at the time were searching and browsing news, and these two accounted for a very large proportion. Therefore, products with a library of songs have inherent adaptability advantages.

With the advent of mobile Internet, mobile phones are not naturally particularly suitable for search scenarios. Secondly, mobile phones have greatly extended the time users spend using the Internet, allowing them to access the Internet anytime and anywhere. The same is true for music apps. In the past, users might listen to music at fixed times, download music from their computers and put it into MP3 players, or listen directly to it by connecting it to their computers.

3. Users want to open the music app and listen to the music they want, rather than just searching

Nowadays, on mobile phones, users may have the desire and need to listen to music anytime and anywhere. At this time, searching alone obviously cannot meet user needs. Users particularly hope to be able to listen to good music directly by opening the APP. Such products will gradually replace music library products and become a mainstream method.

In 2013, all mainstream music products on the market were music library products. Users opened the APP to search rather than share. The only exceptions were products made by startups such as Douban FM, Xiami and Jing.FM. However, their scale is still relatively small, their influence is insufficient, and there are some problems in the polishing and refinement of their products.

How did NetEase Cloud Music build its good reputation?

Generally speaking, a product’s initial positioning has already determined many things that follow, such as direction, development speed, whether the ceiling is low or high, etc. To answer this question, we must start with product positioning.

☞How to do product positioning?

First of all, you need to consider the users, the situation of the entire industry and market, as well as some future development directions, and consider the focus of the product based on these aspects. For users, the most basic task is to find and divide user groups.

1. Slice users

When dividing user groups, you need to choose the factors for the division. For music, we chose two factors: music preference and age .

We visited and interviewed many users at the time and conducted some qualitative research. Through these studies, we found that music preferences and age are closely related to music. Everyone's musical taste changes with age. People of different ages may have different preferences and attitudes towards music, so age is a particularly important factor.

At the same time, in the field of music, different levels of preference represent different user groups. Some are serious audiophiles who cannot live without music; some are people who make a living from music, such as artists, musicians, lyricists, composers, arrangers, performers, agents, etc. For them, music is not only a hobby, but also a job; and some are ordinary users, for whom music is just a kind of companionship.

Due to different levels of music preferences, users have different demands for music. If we combine music preference and age and plot them into a two-dimensional coordinate system, we can create a basic user group division.

At this time, I usually put some players in the industry into the divided user groups to see their approximate market share and distribution.

▲

Music preferences of users using different music apps

For example, in the field of music, taking QQ Music, Kugou Music, Duomi Music, Douban, and Xiami as examples, QQ Music and Kugou Music cover a wider age range, and the level of music preference is relatively low. Users do not have special needs and are only interested in playing the music stored in their mobile phones.

QQ Music is younger than Kugou Music because Kugou Music developed most rapidly during the early PC era of the Internet and accumulated a large number of users. For QQ Music, because of the existence of QQ, many young people go online for the first time and come into contact with the QQ software. Therefore, the age of QQ Music users is relatively younger than that of Kugou Music.

The user groups of Xiami and Douban are somewhat similar. They mainly target older users with higher music preferences, which is what we call senior music fans. Xiami’s age range and music preferences are higher. Xiami’s main feature is a more professional music database. All kinds of music from all over the world can be found on Xiami. This was their positioning at the time.

Douban places more emphasis on the sense of surprise of music discovery, so its age group and music preference level are lower than those of Xiami. Duomi Music is a product that is completely based on mobile devices. The user group it targets is relatively young and their music preferences are higher than those of QQ and Kugou, but lower than those of Douban and Xiami. It just happens to target a relatively blank user group.

The user group was also the best entry point we analyzed at the time, which means that NetEase Music and Duomi Music had similar ideas when it came to product positioning.

We have selected a user group whose music preference level is above QQ Music and Kugou Music, and whose highest level will be Douban Music and Xiami Music, and whose age group will be relatively young.

Why choose a young user group?

Because age is a key factor in forming a user's musical taste, if you target this user when he is relatively young, the user will develop along with the product, and the stickiness will be very strong.

2. Tags and playlists, which one should be used as the core function?

Music content can be organized by tags or playlists, which one would you choose? Label or playlist? This is a classic product selection question.

Product managers face a variety of choices every day, and the choice is related to the user group and product positioning.

Labels have their advantages, and playlists have their advantages. When I do this analysis, I will first consider which type of core users I am. Tags are very convenient for discovering music, even better than playlists. A song can be tagged with multiple tags, and multiple songs can share one tag, which creates an N-by-N mesh structure.

Compared with a list, its network will be more complex and more divergent. However, it lacks the concept of a creator. Putting a label on a song isn't a very creative act. It can only bring different songs together.

Compared to labels, playlists have something more. A playlist is created by one person. What songs to choose and how to arrange them gives this person a lot of value.

When considering product positioning, NetEase Music attaches great importance to discovery and sharing, and therefore pays special attention to UGC. We very much hope to bring value to people, encourage listeners to pay attention to great music talents, accumulate their own reputation, contribute more value to the community, and ultimately form a community network. In this way, the stickiness and activity of the entire community will become higher and higher.

Therefore, we chose to use playlists to organize music content as our core function.

☞ How to bridge the gap?

When doing product positioning, in addition to industry analysis, market analysis, and user group segmentation and analysis, we also consider future development directions. Things that need to be considered include which areas of user base will expand rapidly, what kind of user needs will be suddenly stimulated, and how the product can grasp its positioning in this process, adapt to this change and obtain the greatest benefits and interests from it.

When I was working on mobile Internet music in 2013, I wondered what would happen when everyone had a smartphone and could listen to music anytime and anywhere?

The first is fragmented time. Users want operations on their mobile phones to be as simple as possible, so the traditional editor recommendation mode and music library search mode are not suitable. At this time, users need a large amount of rich content and things that match their interests. Data will be particularly important. We need to obtain users' interest data on music in order to make accurate recommendations to users, while also revitalizing the music library.

1. Meet user needs anytime and anywhere, allowing them to listen to their favorite music at any time

How to activate the huge music library depends on the huge user base of the Internet. Therefore, we emphasize the characteristics of the cloud so that you can access your favorite music anytime and anywhere.

Combining these points, the way users obtain music will change greatly in the next three to five years. They will have very high requirements for the richness and personalization of music content. At the same time, they will also hope that the operation is simple and can meet the needs of listening to music anytime and anywhere.

2. The product cannot be positioned too high-end or niche

▲

The difference between high-end users and ordinary users

If a product is too positioned in a niche and high-end market, and there is no good way to extend it to mass products, it will encounter a very large gap. Many innovative products often encounter the situation where early users are very active but the churn rate is very high, and then the product gradually begins to decline.

NetEase Music and another product Jing.FM were launched at almost the same time. NetEase Music is positioned to satisfy both high-end niche users and mass users, while Jing.FM is more focused on satisfying high-end early users.

The core function of Jing.FM is to describe a current scene by voice or text, and then recommend music. For high-end users, this is an attractive feature. But it is different for general users. General users don’t know how to use this product.

Although Jing.FM later launched rankings to help general users use functions more conveniently, it is difficult to change the product positioning because it has focused on innovators and early users from the beginning.

Although NetEase Cloud Music was initially positioned for high-end users and niche users, it is able to connect high-end users with mass users very well.

For high-end users, it is a great sense of accomplishment when their creations are played, collected, shared, and commented on by others.

For the general users, they can consume things that they could not consume before in this place, and obtain content and unique value, which is also very good for them.

We have established a strong connection between niche users and mass users through some functions, which is very important.

Therefore, when launching a new product project, if you want to enter a relatively narrow market, you must consider whether this part of the users can meet the product model and business model. If not, expansion is necessary. Will we encounter a chasm when expanding? Does this product model work for both niche users and mass users? Can niche users and mass users coexist in the same product? These are all issues that need to be considered.

☞ What effective methods have we used in the process of building a good reputation?

1. Playlist

The first is the playlist. The playlist format was not invented by NetEase Music. Why haven’t the playlist functions of other products become popular?

There are four key elements here.

The first is simplicity. The playlist function of NetEase Music is very simple and has the lowest threshold.

Secondly, it is the core function of NetEase Music, and most users will come into contact with the playlist function.

The third point is connection. We connect some creative opinion leaders and some general users through playlists. There is a good bond between them through the playlist.

The last word is temperament. Cloud Music is essentially a community product, and every community has its own atmosphere and temperament. The user positioning we chose is the middle segment of the region, to establish an inclusive community relationship.

2. Music community

The second is the music community we have created. The most critical design concept of a community is to govern by doing nothing and let users play by themselves.

Good products and good communities should only need to give users a point and then rotate on their own. Such communities have the strongest vitality. The most notable feature of the Cloud Music community is music reviews, which are the greatest charm of the community. This function is very simple, one comment plus one like and one reply.

3. Create a comment function

We have changed the music listening habits of many users through music reviews. Why are the comments on NetEase Music so popular? Here we can see three very typical comments. The first one is about "Charlotte's Troubles", a song that resonated with users in Beijing; the second one is about a Korean group, and the comments their fans made under their song. The third one is some jokes, which can often be seen under folk songs. These three types of comments basically represent the reasons why NetEase Music users like comments so much.

Comments are actually just comments and likes, which is a very simple function. We hope that users will not only listen to songs, but also regard this as a community, constantly consume these contents, and continue to use it.

Penguin Intelligence once released a report that only 5% of people would read comments while listening to music, which is a very niche demand. But now 30% of people read music reviews. For NetEase Music, we want to create a community about music so that people who love music can have a great time in this community. On NetEase Music, nearly half of the users will read comments.

How to explore the demand for comments?

First, we observe user behavior and dig out valuable points in Weibo, Tieba, and APP. Secondly, we observed the comments of NetEase Music users in the APP and summarized the characteristics.

Analyze users’ psychology from the above two points. There is a very strong connection between music and emotion. When users listen to music, they often express their emotional demands, and these demands will resonate with each other. If a user writes down these resonances, other users who have similar experiences or similar feelings will have a strong emotional response. They will like and spread it. In this way, the comments will gradually become popular.

One of the biggest benefits of NetEase Music’s comments is that they change users’ existing habits. So are NetEase Music’s comments and NetEase News’ comments the same thing?

For news, immediacy and effectiveness are very strong. If a piece of news has passed today, the possibility of you reading it will be very low, because it lacks consciousness sedimentation and accumulation. But music reviews are different. A song, even one from ten years ago, is still listened to by people now, so its strongest feeling is resonance. Different people, listening to this song at different times, have the same feelings and emotions, which can trigger a very strong resonance.

Comments on NetEase Music are more able to express an individual's feelings, so we like the comments instead of posting a lot of replies.

The second thing we changed was the user habits. Previously, users would put the app in their pockets when listening to music, and they did not spend much time looking at the app. The demand for advertising was also weak, and it seemed that the revenue of the entire product could not be generated. We have changed the user's habit from putting the APP in their pocket when listening to music to reading comments while listening to music. In this way, the entire usage market and stickiness will be strengthened, and the commercial value of the APP will become higher.

4. Personalized recommendations

The fourth feature that creates a good reputation among users is personalized recommendations. Personalization is the future development trend. Making users extremely lazy is the music recommendation effect we pursue at NetEase Music.

Although NetEase Music is not the first to offer personalized recommendations, we are the best product in the entire music industry in terms of personalized recommendations. Let users get in touch with and use the product functions as much as possible, and promote it.

Now basically all musicians in the country can accept the concept of personalized recommendations. We previously conducted a questionnaire survey among musicians who have a very high rate of usage of personalized recommendations. We found that their usage rate of personalized recommendations reached 75%, and 50% were willing to recommend this feature to other friends, which is a very large proportion.

(1) Create a sense of surprise for users

Personalized recommendation is actually a long-term optimization process. In addition to working hard on the algorithm, what other role can product managers play in this process? Let me give you a simple example and look for some surprises in recommendations from the perspective of studying user psychology.

How to make recommendations surprising is a very important issue in personalized recommendations. If the recommendation is too ordinary and fails to surprise the user, the user will not particularly like the recommendation. Once the user feels surprised, he will like the personalized recommendation function very much and then form a good reputation. Therefore, we place great emphasis on surprise when making personalized recommendations.

The process of developing this function is a process of studying user psychology. By observing the music reviews posted by users, we find some music through the recommendation algorithm. We seized this feature and extracted some text for analysis, taking some music as seed music, and continuously diverged and expanded it, so as to find the music that can bring the most surprises to users.

In addition, we also want to explore users' various interests. Because users' interests are often not single, a person may have a variety of interests in life, and they may be connected to each other to form a rich person.

For example, books, movies, and music are all the user's interests. If we can understand the user's interests and preferences in books and movies, we can recommend music that he may like. We read some movie-watching records shared by users on Weibo, and recommended the original soundtrack of the song in the Cloud Music APP. The result was very good, which gave us a strong sense of surprise.

Another aspect of personalized recommendations is that, generally speaking, for products that purely provide personalized recommendations, the recommendation algorithms will become narrower and narrower.

(2) UGC and recommendation algorithms work together

▲

UGC and recommendation algorithms work together

NetEase Music has done a lot in this regard. We found that the entire music system is a combination of UGC playlists, user attention, and recommendation algorithms. When the recommendation algorithm falls into a loop, the user discovers something new in the playlist or dynamics. At this time, the recommendation algorithm can immediately capture it and expand the scope of the recommendation.

In order to allow users to discover more surprises, we will constantly consider how to discover less popular songs.

▲

User and Music Pyramid Model

We divide users and music into two pyramids. The user pyramid is a positive pyramid, that is, the highest level of users are positioned as connoisseurs, and this part of users may only account for 1% of the total user base. They like some fresh music and pursue high-end and showy feeling. We recommend personalized music to these users. Through their ears, we pick out the better music, which we call bed music, and it accounts for about 80% of the entire music library.

By picking out good music through the ears of connoisseurs and sharing it, it can reach the ears of those UGC content creators and disseminators. These users are very good at making playlists. They can organize music into playlists and recommend them to more users to share. Such music will gradually become popular music, which accounts for about 20%.

Through our algorithms and ecosystem, we provide more mainstream users with the tools to vote through playback, sharing, collection, liking, and commenting, ultimately selecting the most popular music in the product, which accounts for approximately 1%.

During the whole process, a model that combines users and music allows some niche and unpopular music in the music library to be exposed, gradually producing the most user-friendly music in this product.

Compared with the traditional editor selection model, this has a huge advantage, because it is produced through repeated selection from users' behaviors, and it is not just the hot songs on the charts, it also includes many unpopular songs. You can find a lot of music on NetEase Music’s rankings that is rarely seen on the rankings of other products. This is the charm of this model.