In her two years at Tencent, she learned these 15 content management tips

Two years ago, I graduated with a master's degree from Renmin University and accidentally entered the mobile Internet industry. After coming to Tencent, I was in the Tencent Mobile department. At that time, I was not very interested in portals. When I learned that our group was engaged in the push operation business of the mobile QQ browser , I was very happy to participate in it.

In two years, I have seen the number of daily active users of QQ Browser more than double, and the push DAU I participated in has also more than doubled. In the past two years, I had times when I just started working and had no idea where to start. I was so stressed that I went home and cried late at night. There were also times when I was alone in the company looking at data late at night and was excited about discovering a problem or making progress. There were also times on weekends when I talked with my colleagues about competitive product strategies until the early morning, discussing feasible plans for the future of the product. What impressed me most was that when our product became the number one in the industry, we toasted each other at the celebration banquet.

Looking back, I am grateful along the way. I would like to thank Tencent for providing this big stage, thank the product itself for being excellent, thank the excellent colleagues in the team including product development, editing and design, and I would also like to thank myself for being the first one to arrive at the company and the last to leave, and for pondering data even on weekends and late at night.

These past two days I have been thinking about what I have gained in the past two years, and finally summarized these 15 pieces of information , including 5 preparations to be made in the early stage + 3 tools to be used in content operation + 4 processes of content operation + 3 capabilities that need to be possessed . Without further ado, let’s get straight to the point: how to do content operations well.

Before the official operation, five aspects of preparation need to be made

1. Understanding of your own products

First of all, you need to clarify what your product positioning is, what user needs it can solve, in what scenarios users will use your products, what is the position of your products in the industry, and what your business model is. What are the future development directions of our own products? Only with a deep understanding of these can operations be done better.

For example, when I first started to work on the push of QQ Browser, QQ Browser was still positioned as a mobile Internet browser, and its main services included information, novels, videos, searches, etc. It can be seen that information is just one of many services, so at that time, our understanding of push was that although we did information push, we were a mobile browser after all, and it was not suitable to push too much information content to users.

Later, our product positioning began to transform. Changes took place in the industry, and browsers were transforming into information readers with personalized recommendations. We also launched personalized recommended information feeds in version 6.4. It can be understood that the proportion of information in our product positioning has increased in importance, which has put forward new requirements for our operational work. We must not only push more information content, but also go further in depth and breadth than before.

In short, on the basis of in-depth understanding of one's own products, one must promptly know changes in the industry, changes in one's own product positioning and services, and operational strategies must also keep up with changes in the industry and one's own products .

2. Understanding of the users you operate

Before you start your formal work, you can take a look at a basic user portrait. When I first started doing the job, I didn’t know the users very well and I always felt at a loss as I didn’t know much about them. After working for a while and having a certain understanding of them, I became more adept at it. If you have a clear understanding of the user from the beginning, work will go more smoothly.

User content that needs to be understood: basic user attributes, such as age, gender, region, device model, education level, Internet habits, and content preferences.

How to quickly understand users? I think there are several methods: ➀ Basic user portrait . Sometimes the user research department will have it. Find one and come and have a look. This is the simplest and most efficient way. ➁ Basic data . You can get a rough idea of ​​the data from the background. You will gradually understand the specific data as the operation rhythm deepens. Data is God's handwriting, and you can gain insight into users through data. ➂ User interviews . The early stage of user interviews is to contact users. At the beginning, you don’t know where the real users are. Later, I found that there is a particularly good opportunity to contact users, and that is when users complain. Sometimes users send complaints through product complaint channels , and sometimes they complain on Weibo. In short, when I receive a user complaint, I will chat with the user through QQ or Weibo private messages. Asking him the reason for his complaint is one aspect, and the real purpose is to have a more in-depth chat so that I can better understand the user.

3. Current status of product data

When you first take over a product, you will know the product’s current daily active users, conversion rate , and total users. This is the data for the first step.

The next step is to look at the operational data. Take push as an example. What we need to look at are: how many pushes are sent every day, when each push is sent, what the content of each push is, what the click-through rate is, how many pushes are sent, and how many users click on them.

4. Core KPIs and calculation formulas for operations

This place understands the core KPI of operations, which is actually the goal. For operations, the most common KPI is daily active users. Operations colleagues should always keep DAU in mind. The first thing they should do when they come to the company every day is to open their computers to check DAU, to see whether it has increased or decreased since yesterday, and how far they are from the KPI.

The second thing is to understand the calculation formula of KPI. I found in my work that many people don’t quite understand the KPI calculation formula. The common KPI calculation formula for operations is DAU = total users * conversion rate.

For content operations, the common calculation formula is clicking users = display users * conversion rate. In the calculation formula, the numerator and denominator of each place must be made clear, and the funnel model of the operation link and the data on each funnel must be made clear. The KPI is clear, and the KPI calculation formula is also clear. The next thing to do is to break down the KPI. Based on the KPI and its calculation formula, break down the feasible ways to complete the KPI, and then execute them accordingly.

5. Competitive products, strategies and data

Products generally have competitors, and the competitors are different for products at different stages of development. If you want to get started quickly at work, first find your competitors. In this field, there are competitors that are the best in the industry. You should focus on competitors that are comparable to you, and also competitors that are not as good as you.

There is no doubt that there is a lot you need to learn from competitors that do better than you. Competitors that perform similarly to you may even be worse than you. In fact, their data may not be as good as yours, but that is not necessarily because their operational skills are worse than yours. They also have their good points that you can learn from. Even if you don't do well, you can still take a look at it. When you see a virtuous person, you should try to emulate him; when you see an unvirtuous person, you should reflect on yourself.

When looking at competitors, the first thing is to find the competitors, the second is to understand the competitors' operating strategies, and finally it is best to understand the various conversion data of the competitors. Now that you have chosen a field, you should immerse yourself in it and find the competitors in this field and the people behind them.

When I was doing push operations in QQ Browser, my biggest feeling was that doing a good job requires continuous learning. Because our work changes quickly, sometimes there is not much learning material. Apart from exploring on your own, the most effective way is to learn from your competitors. When you are really pushed to the limit, you will take the initiative to get to know many people from competitors. Sometimes you will even become good friends with your competitors after chatting with them, which is beneficial to both parties.

Getting Started: Three Tools for Good Operations

At work, I use the following three tools almost every day and have benefited a lot from them. When it was time to hand over my work, I compiled a work handover document. When I opened the folder, I found the following three tools: a mind map, an Excel spreadsheet, and a Word document.

1. Mind Map

I put mind mapping first because I think it is the most important. I usually use mind maps to write strategies.

After clarifying the KPI and understanding the calculation formula of KPI, I will draw out the factors that affect KPI and from which aspects I can improve KPI. This requires structured thinking. First, I will look at the data and talk with competitors, and write a part of it. After reaching a certain conclusion, I will discuss with my friends to see if those strategies are feasible and whether my friends have better strategies. After reaching a consensus, I will draw it with a mind map, which will make it convenient for me to advance step by step in the next work by referring to the mind map.

2. Excel table

When I was organizing my handover documents, I found that Excel spreadsheets were the most numerous. In my daily work, the thing I deal with the most is Excel. Excel spreadsheets are generally used to do the following:

➀Basic data recording. Here are the general data. There is a table showing the annual DAU, which contains the average DAU for each month. There is also a table showing the 12-month DAU, which includes the average for each month. Each month's table also contains the daily DAU. In this way, whether the boss asks or you yourself, you will know clearly the development direction of the products you are operating.

➁ Refine data records. The data just recorded was just DAU, which has a very coarse granularity. Next, some detailed data will be recorded, for example, the number of users who clicked on the push during the morning period, the average value last month, and the average value last week. If we find that the data for one day is below or above the mean, we will check the reason. Based on the data here, we can adjust our product strategy. For example, the morning click-through rate reached a certain bottleneck and has remained that way without changing. We may make changes by adjusting the delivery time or the delivered materials to see if the click-through rate changes.

By the way, a good operation is one that is planned in advance, executed as planned during the process, and results are obtained afterwards, accumulating experience for the next operation. Looking at detailed data is the process of verifying operational results.

➂Specific copywriting data. The data here is even more granular, down to the click-through rate of a particular piece of content. By summarizing this type of data regularly, we can gain a deep understanding of users' content tastes. I think such summaries can be read every day, but if you really want to sit down and do analysis, you need a larger data sample size. So we usually summarize once a week and once a month.

For example, we will look at which of the selected content has the highest click-through rate, record it, and try similar topics again in the future. There are also some with low click-through rates. For example, a major earthquake occurred in a trench in South America. After this push is sent, we look at the data and find that the click-through rate is particularly low. In the future, we will reduce the number and volume of such news releases.

You can also compare the data horizontally.

You should summarize more after doing things. If you don’t summarize, things are just cheap experiences. The summarized experience can be reused in future work.

➃Operation diary. As I understand it, there is no fixed template for an operations diary. Just record it based on your product characteristics and personal style. I record all the product optimization strategies and operational changes I made that day on a daily basis, and write notes at the end. I usually check the next day to see what data this strategy has brought.

Operation is a meticulous job and a long-term battle. Sometimes the improvement of data is not brought about by a certain work. Many factors will affect the improvement of data. Our job is to find all the factors that may improve the data in the early stage, and then try them one by one in actual work. Some factors can really improve the data, while some cannot. We need to record which ones we tried, which ones we did not try, which ones have been verified to be effective, and which ones have been verified to be invalid.

In particular, special dates when some major functions are launched and have a greater impact on the product must be recorded. After I started working, I found that both my colleagues and leaders would often ask me about some data, including the dates corresponding to these data. If I didn’t keep an operation diary, they would have no way to check when they asked. I kept an operations diary, and when my colleagues asked me when a certain strategy was started, I could quickly give the answer by checking the operations diary.

Another function of the operation diary is to remind yourself to change quickly. The Internet industry is developing rapidly, with rapid trial and error and rapid iteration. In terms of operations, our optimization cycle is measured in days. Because the industry and users are changing, if we change slowly, it will show in the data. So, we have to keep a fast pace of change every day. During the phase when our operating indicators were growing fastest, our change cycle was days. When writing an operations diary, it also reminds us: You should change today. If there is nothing to write in your operations diary one day, it is time to reflect on the optimization rhythm.

➄ To-do list. After listing the basic strategies for completing KPIs in the previous mind map, and after looking at the data and understanding the problem, the next step is to schedule execution. I usually keep my to-do list and operations diary together. I usually look at my to-do list twice a day. First, when I get home in the morning, I open it to see what goals I need to accomplish that day. Then, I look at it again after get off work to see how much I have accomplished today.

Operations is a process of clarifying goals, breaking down goals, quickly executing, verifying conclusions, and applying conclusions. Mind mapping solves the process of clarifying and breaking down goals, while Excel spreadsheets solve the process of rapid execution and conclusion verification. To verify the conclusions, we need the third killer tool - Word documents.

3. Word Document

I use Word documents to do two things. The first is to write operating specifications; the second is to write experience summaries. Operational specifications generally state what should and should not be done; experience summaries state how to do things better. The above content all comes from practice and data. After doing more, sort out what you have done, look at the data after finishing, see how to do better, and then write it all down.

First, let’s talk about operating regulations. I think the most important feature of operating specifications is that they are always being optimized and iterated. They can only be named by dates and versions, and they will never be perfect. In actual work, we implement the specifications, but we also need to see what imperfections there are in the specifications themselves. Many times, operating specifications cannot keep up with actual changes. When the operating method changes, the standards should also change.

What kind of standards are good standards? I think it is detailed and executable. For example, when we make push specifications for major breaking news, take earthquakes as an example. We clearly specify what kind of push can be issued for earthquakes of different magnitudes, and even specify which cities and regions have earthquakes of certain magnitudes that can result in what kind of push being issued. The output of the specification should be unambiguous and not contain too much ambiguity, so as not to make people dizzy and confused when executing it.

Another is experience summary. The power brought by an experience summary can last for a long time, and experience summary is also a very interesting thing. When we were doing push, we summarized many times what kind of topics users liked the most, what kind of copywriting users liked the most, and what kind of punctuation users liked the most.

Experience summary is a sublimation of the tedious daily operational work. A good summary can improve the operational efficiency in the next stage. Output a written summary of experience at least once a month.

Formal stage: How to do content operation

What is content marketing? Zhang Liang said that content operation refers to the creation, editing, organization, and presentation of website or product content, thereby improving the content value of Internet products and creating content that has a certain promoting effect on user stickiness and activity. In my understanding, content operation is the process of matching appropriate content to appropriate users and satisfying user needs through content.

1. Where does the content come from?

To solve the problem of where the content comes from, we must first clarify the content positioning. Content positioning is based on product positioning and user positioning. On this basis, user needs are analyzed and then content positioning is derived.

For example, I used to run a WeChat public account called "Qian Yan Wan Yu". I positioned it as the "spiritual home of media professionals". The target users were media professionals, and the content was positioned as articles that were helpful to media professionals. So, for a while, we posted articles about how to improve content production capabilities, whether to leave traditional media, and how traditional media should do a good job in new media . These articles had a high click-through rate and met the needs of users at the time.

Clarify the content positioning, and then find the best articles that match the content positioning. The first step should be to find the content source. Content production can come from the following places: content produced by traditional media, content produced by professional journals and books, content produced on WeChat public accounts , content produced on Zhihu, content produced in other forum communities, content produced by KOLs , original content produced by oneself, etc.

In fact, when it comes down to it, we need to grasp the source. For example, when we make personalized recommendations for push, there is a content label called "Gender Emotions". After user analysis, we found that most of the users in the emotion label are male, so we will publish some articles that talk about gender emotions from a male perspective. We found that there is a public account called "Tips for Picking Up Girls". The content produced by this account is particularly popular with our users, so we just focus on this account. We do this with every label.

Determine the field, find well-known KOLs in the field, find where the KOLs' content is presented, and then gather these high-quality content in your own hands through crawling, negotiating cooperation, etc. When I was working on Qianyanwanyu, I harassed all the KOLs in the media field who could produce high-quality content, and they all became my WeChat friends. High-quality content can be guaranteed by solving the content source problem.

2. How to select content?

Content selection is the process of selecting high-quality content that matches user needs from the many content sources previously captured. What kind of content is good content? Simply put, there are two criteria, one is that the content is interesting, and the other is that the content is useful. Either it satisfies the user's need for relaxation and pleasure, or it satisfies the user's need for knowledge.

There are also two ways to select content. The first is manual judgment. Just read the article quickly and judge its quality. This place relies on experience. Generally, when you first start doing something, you can’t go too fast. You have to read carefully. Only by reading carefully can you ensure the content operator’s ability to control the content in this field. When I first started working on Qianyanwanyu, I had to read 10 articles about the media field a day before I could select the best one. After you become familiar with it, you can judge the quality by quickly scanning the article.

In addition to judging the quality of the content, manual judgment is more importantly about judging the selection of materials itself. What is being called upon here is past experience, so it is particularly important to summarize what has been done. One of our push tags is "anecdotes". We summarized the data and found that the click-through rate will be particularly high when it comes to things like "big fish", "big snake", "aliens", "corpses", "mummies", "lottery", and "strange diseases". Whenever we come across such topics, we will pay special attention to them.

The second criterion is data judgment. It is to judge the quality of the article through data indicators, such as the number of clicks , click-through rate, number of comments, etc. For example, when we are doing push, we need to grasp the hot events of the day. How do we judge the hot events? We judge through data. First, we look at our internal article rankings , half-hour traffic rankings, and five-minute traffic rankings. Then we look at external data, the number of comments on an article on major clients, and other websites, such as the Baidu Hot List and Sina Weibo Hot Topic List. Determine the hot spots of the day through various lists.

After many comparisons, it is difficult to judge which one is better, manual judgment or data judgment. In most cases, both are needed. Many times, when an article is first published, the number of readers is low. Then we start to operate it and it immediately becomes popular. Other competing products also follow suit. At this time, we rely on human judgment. Sometimes it’s hard to tell how good an article is manually, but it has a high readership. We have also operated it, and the data is indeed very good. The best approach is to combine human judgment with data judgment.

Data judgment is actually CTR- oriented, and sometimes CTR-oriented content does not mean good quality. For example, the Lujiazui incident was particularly popular some time ago, but this incident was a bit low-level, and private matters were spread on the Internet, so it was not worth paying attention to. There is also an account in the field of media that posted an article, writing about this matter from the perspective of media. Although it has a large readership, I don’t think it has much value. It is always shameful to make a fuss about the private lives of ordinary people. Posting such articles does not do much good for the brand or deepening user impression.

3. Processing the content

Even if the content is selected, it is not suitable to be pushed or presented to users immediately. It needs to be processed based on understanding of the users. When I was working on Qianyanwanyu before, I would change the title and write a summary. Later, when I was working on push, I still changed the title and wrote the copy.

There are many such cases . For example, the copy written according to news materials: 3 people's bank cards were stolen and 450,000 yuan was transferred to the same person. The bank said there was no abnormality in the system - the click rate was 5.11%; the copy rewritten from the user's perspective:

3 people’s bank cards were stolen and 450,000 yuan was spent without leaving their bodies. They didn’t tell anyone their passwords, so how could the money be stolen? - Click rate 6.47%.

We have summarized this issue many times and written many articles on this subject. I won’t say more here. Common ones include "copy ending with a question has a higher click-through rate", "copy that is half-covered and half-revealed has a higher click-through rate", "titles with numbers have a higher click-through rate", "titles with sharp contrast have a higher click-through rate", etc.

How can I write better copy? You can try it with the help of abtest. Sometimes we will write five or six pieces of copy for a piece of news to conduct ab test, and finally analyze the data to summarize the rules. The high click-through rate can reach 10%, and the low click-through rate can reach 2%. There’s a saying that goes “ If your copywriting isn’t good enough, it’s probably because you’re not writing enough .”

There are so many common ways to write copy. Just look at the titles of articles with high traffic on WeChat public accounts, and you will roughly understand how to write a title. However, deliberate sensationalism, malicious clickbait , and awkward guidance of user clicks are all negative examples and should be avoided as much as possible.

4. Distribute content to users

There are also many factors that affect the efficiency of content operations. Simply put, they are reach and display. The points involved in arrival include: When will it reach the user? What kind of users are reached? What is the reach rate? The points involved in the display include: Where to display? How is the display format? How long is the show?

Let’s talk about arrival first: the first factor that affects is time . I think content operation is to push the right content to the right users at the right time. When is the best time to push click-through rate? The golden time is different for different products, and it is also different for different content. For example, when we were doing push operations, we found that when we pushed current affairs news at seven or eight in the morning, the click-through rate was not high. At this time, the click-through rates of life encyclopedias and inspirational stories were higher. After 10 a.m., the click-through rate of current affairs news increased.

Any subject matter should have a golden time point for pushing, and this golden time point can only be found slowly.

The second factor is the user . In the formula for click-through rate, click-through rate = users who click divided by users who send messages. If you want to increase the click-through rate, you need to increase the number of users who click and reduce the number of users who send. This is where refined operations on users are needed. For example, when we selected users who were interested in military equipment, we selected a relatively typical military equipment material and sent it to non- female users to see the effect.

The third factor is reach . Here, products and technologies need to work together to solve the problem of reach. There are also many influences from the users themselves, which I won’t go into detail here, but this is indeed a point for optimization.

Look at the placement . Of course, the better the display position and the longer the display time, the higher the click-through rate. But how to combine different types of materials to maximize the operation effect requires adjustments based on the content.

Finally, the display form . When running a WeChat public account, the title, pictures, and abstracts affect the display, among which pictures should be especially good. Later, when we were doing push, we found that adding a small icon to the push could also increase the click-through rate. I heard that adding emoticons to push will also increase the click-through rate.

The above is probably the basic process of content operation. Finally, regarding how to do operations better, I think the following abilities also need to be improved.

1. Learning ability

My biggest feeling after working is that if you want to do your job well, it is not enough to rely solely on previous knowledge. You have to keep learning to keep up with this rapidly developing industry. There are two paths to learning. One is to read by yourself, such as books and articles. The second is to talk to people. Talking with peers and colleagues is part of the learning process.

I set a KPI for myself, to read at least a few books every month and to have in-depth exchanges with a few friends or colleagues. Here are some recommended books: Internet work ( Growth Hacker , Operation from Scratch, User Power : Demand-Driven Products, Operations and Business Models, Everyone is a Product Manager , Big Data Intelligence, Machine Learning, Newbies Can Do Data Analysis ); Work efficiency (One Step Forward, The Seven Habits of Highly Effective People, Dark Time, Treat Time as a Friend, Illustrated Pomodoro Technique, Self-Control, Willpower: The Psychology of Focus, Self-Control and Efficiency, Reshaping Work, Energy Management); Excellent business operations (Built to Last, One Win and Nine Losses: The Secret of Uniqlo's Global Popularity, Letting Go of Success for a Day, The Hard Thing About Entrepreneurship , From 0 to 1: The Secret to Unlocking the Future of Business, and The Biography of Chu Shijian)

2. Teamwork Ability

Operations colleagues will interact with partners in other trades at work. If they rely solely on operations, KPIs will definitely not be achieved. They will need support from development, product, and design partners.

A friend asked how operations students should interact with developers. My understanding is that operations and development are not in an adversarial relationship, but a cooperative relationship. The premise of cooperation is common interests. Before looking for a developer, first make sure you share the business KPI with the developer. If yes, think about how much your needs contribute to the KPI completion. If so, think about whether the technical implementation cost is high. Although we would say that "any problem can be solved through technology", the technical cost of implementing some needs is too high, but the benefits are not great. The priority of such requirements will be downgraded by one level, so there is no need to mention it to the developers for now.

Finally, think from the developer's perspective, what benefits will this requirement bring to him, and communicate in a language that the developer can understand. Don't think that your developers are too omnipotent, and be appropriately considerate of them. If you make some requests that contribute greatly to the completion of KPIs, have low technical costs and high returns, I think no normal developer will refuse.

In order to communicate better with teammates, I think we should fully respect each other's professionalism and fully trust our teammates. At the same time, we should also understand their work and thinking. For example, we can read programming or design books.

3. Business reflection and summary ability

After work, I learned a word from my colleagues: “review”. My understanding is that we should think about what we have done in the past again, step out of the mindset of those involved, and reflect from the perspective of an outsider to see what we did well and what we did not do well. Whether it is good or bad, it should be summarized.

A good memory is not as good as a good pen. The conclusions reached should be put into writing. Looking back now, I feel like I have gained something new every time I read the diary I wrote in the past two years.

Write down what you did well so that you will know how to do better next time. Pay more attention to things that are not done well or even done wrong. I believe that any operational accident is not accidental, but inevitable. It simply exposes problems that have accumulated over a long period of time. For every mistake, we should reflect on the root cause, conduct special research on the weaknesses, and find the most comprehensive countermeasures to solve them.

Finally, if I were to use three sentences to summarize my work experience over the past two years, they would be: There will always be many problems encountered in the past two years, but these problems will eventually be solved; if someone is complaining about your product or business, it means it is a good thing, as there are opportunities hidden; to complete a task, there is always another plan B with lower cost and better effect, and plan B is always a secret. What we need to do is to find the more perfect and deeply hidden plan B.

I also want to give a message to my future self: completing KPI is far from enough, what is more important is to exceed KPI.

Source: Huo Qian