Practice: How to use data mining to convince your client?

This article is written at a milestone moment.

On the fifth anniversary of the opening of WeChat official accounts , Tencent released its second quarter 2017 financial report, in which it announced that the combined monthly active accounts of WeChat and Wechat reached 963 million, an increase of 19.5% over the same period last year.

But even with this huge traffic entrance, we need to face a reality - we have entered the second half of new media (WeChat official accounts to be precise). The bonus era of "as long as you open a public account, or register a domain name and do optimization, you can gain hundreds of users' attention in a day without posting any articles" has passed.

But even so, there are still many personal self-media and corporate official media that are catching up. If your public account is not doing well, please don’t blame yourself for being "born at the wrong time." Nothing will always move upward, spiral development is inevitable, but have you really worked hard? Are you working hard in the right way?

Today, I would like to draw everyone’s attention to data collection and analysis when operating public accounts from the perspective of WeChat backend data. I would also like to share with you some of my small ideas and discuss with you.

The following is the framework of the following

1. Data analysis and peer research results

2. The significance and value of data support to operations

3. Limitations of viewing and collecting WeChat backend data

4. Gain insights into operational issues through WeChat backend data

1. User research of peers

When some of my colleagues in the same industry were communicating with me before, I found that most of them were doing public account operations (whether they were self-media operations or operations for the client). When making plans and various plans, they were more likely to "go by guesswork" or "go by feeling". Even if they could achieve good results, they would find that they would always encounter difficulties when copying the methodology, and the experience would be difficult to copy if the circumstances were different.

Most operations partners (who tend to have a liberal arts mindset) do not pay enough attention to data, so they don’t have the habit of collecting and analyzing data. For example, the following figure:

There is another type of friends who value data but are not sensitive to it and do not know how to use it.

“Crawling data manually is exhausting~”

“Where can I find historical data?”

"The client wants a report every minute, but there is too much data, it's dizzying and I don't know where to start analyzing~"

“Why does the total number of readings in the background not correspond to the sum of each reading source?”

I once spent some time to do a small survey of nearly 100 colleagues around me. Here are some of the results I would like to share with you:

Although more than 50% of people think they are familiar with the data indicators of WeChat backend, communication and communication found that the data they look at most are:

1. "New users" and "Total number of users" on the homepage
2. Total number of readings of the picture and text

As for other more in-depth segmented data, it is more difficult to find. I tried using the following test question in my questionnaire (the correct answer is: User Analysis – User Attributes), and concluded that less than 40% of people answered correctly.

2. The Importance of Data Analysis

In the era of mobile Internet , users pay attention to many and scattered things, but their time is limited. We are not only competing for traffic resources with 20 million public accounts, but also competing for traffic resources with platforms other than WeChat (such as Youku/ Taobao , etc.), trying every means to retain users' time.

Whether you are running a personal account or a corporate official media, you have your own purpose, positioning and plan when creating the account. The reason why users follow your public account is that your public account is valuable to them, such as entertainment gossip and funny things that make people relax, learning content that helps people improve themselves, and quick services such as inquiry, processing and purchase.

In my opinion, the premise for achieving success in “ attracting new users , retaining users , and promoting activation” is that you treat the users of your official account as people and friends, rather than fans. What are these users (friends) like? What do you like? What do you want? Do you really understand? The necessity and importance of data analysis are reflected in:

• User analysis: Are your users accurate?
• Graphical analysis: What do your users like? How much do you like it? How valuable is the content you provide?
• Menu/message analysis/activity analysis: Are your users active? How valuable is the service you provide? etc.
• Other analyses will not be elaborated for now.

3. Limitations of WeChat Backend Data

The above figure is a summary of the current status of WeChat backend data, which causes the following pain points for operators in data viewing and collection:

• The related data are not presented in a centralized and intuitive manner, and switching between them is cumbersome and time-consuming.
• The collection of some data can only be done by manual copying, and summarizing is too cumbersome.
• It is not possible to export the data tables of multiple modules for a certain time period at the same time, and the summary is too cumbersome.
• Some data are updated in real time, and it is impossible to monitor the incremental fluctuations at the minute or hour level in a timely manner.
• The data distribution of a single picture or text is too scattered, resulting in the numerical values ​​not matching when adding up the data.
• other.

When WeChat's official data collection and processing became too difficult to use, a number of excellent third-party data products flooded into the industry for operators to use. I selected several good products and made a comparison of the experience. You can use these tools flexibly according to your needs to help you collect and summarize the data you want.

4. Insights into WeChat Data Problems

However, collecting data is only the first step. Once you have the data, how can you use the data to gain insights into the problems you encounter when operating your official account?

Today I will share a small part (the rest will be added later): When you look at the various data indicators of WeChat, which tools can assist you in collecting them, and you should gain insight into the problems that arise in your operational work and seek solutions.

Account overall status and user management list:

It is recommended that operators develop the habit of conducting minute-level monitoring when tweeting or conducting activities.

1. From the perspective of content operation , every time you post a tweet, you can promptly understand the impact of the tweet and activity on the number of followers.
2. From the perspective of event operation , many events are pushed in the form of pictures and texts on official WeChat accounts, and come with rules to guide users to follow the official accounts. Based on this, we can timely understand the incremental impact of the event on users, and adjust the event background configuration in a timely manner based on the event PV/UV.

User Analysis - User Growth

1. From the perspective of content operation, by comparing the ratios of user sources, we can understand the contribution rate of pictures and texts to increasing fans. (Because many pictures and texts will attract new users by means of "introduction" or QR codes in the text; you can learn about the reputation of the public account through "public account search" and "business card sharing".)

2. From the perspective of event operation, by comparing the ratios of user sources, we can understand the contribution rate of the event to the increase in fans. (Because many plug-in activities require users to pay attention before they can participate)

3. Daily user growth report can be used to understand the patterns of user growth (unfollowing, new additions) and guide content and activity operations.

User Analysis - User Attributes

User attributes help to gain insights into user profiles . By tracking the incremental changes in users of different genders, geographical distributions, and terminal distributions, and conducting a comparative analysis of the proportions, we can understand the degree of consistency between the user portrait and the public account user population positioning. When the user proportion of a certain dimension shows an abnormal change (such as a sudden increase in users in Foshan City, etc.), it is necessary to find out the problem in a timely manner.

Real-time data of pictures and texts in mass distribution

It is recommended here to perform minute-level monitoring (currently Xigua Data , Xinbang , and BLUEMC support this service, but it requires payment to use it), which can be combined with minute-level trend monitoring of user increments and activity PV/UV for analysis.

1. From the perspective of content operations, we can understand the popularity of the post-tweet dissemination and user interaction, and make timely adjustments to use other channels (store clerks, KOLs , etc.) to increase the dissemination of articles.
2. From the perspective of event operation, we can deduce the impact of image and text opening rate on event participation conversion rate , and provide a basis for adjusting the background configuration of the event process.

Graphics and text analysis - single graphic and text

1. Export all indicator data of a single picture or text in this dimension (currently Yiban supports free export) and conduct analysis to effectively guide content optimization.

(1) For example, if the ratio of “public account conversation reading” in the reading source is high, it means that the title is eye-catching; otherwise, it means that the title needs to be optimized. In addition, this indicator does not mean that the content of the article attracts users, it may be a clickbait title .

(2) If the sharing rate is high, and the ratio of "reading in Moments " and "reading from friends' forwarding" in the reading sources is high, then it is possible that the content of the article is recognized by users, thus triggering self-propagation by users, and the title may be eye-catching, increasing the secondary opening rate; but it is also possible that the forwarding is a political task (such as the headquarters requiring its stores and employees to forward it). In addition, a high sharing rate but a low secondary opening rate is closely related to the title of the picture and text as well as the language used when forwarding.

(3) The number of WeChat collections can more truly reflect the usefulness of an article to users.

(4) Reading the original text page can reflect the transformation effect of the introduction of the main text on the "read the original text" part.

2. For the images and texts pushed during a certain period of time or since the creation of the official account, export all indicator data based on this dimension (currently Yiban supports free export), and classify the image and text content by type, and then compare them (make them into cell color levels or other visual trend charts, the comparison will be more intuitive). From different detailed indicators, you can review the previous content operation results and make a new round of image and text planning.

1. Update and adjustment of graphic column
2. Title Optimization
3. Posting time adjustment
4. Adjustment of the introductory words in the text of the graphic
5. other

Graphics and text analysis——All graphics and text

WeChat has counted the total of all graphic and text data of the entire official account. The data in this dimension can provide a glimpse of the graphic and text effects of the official account since its inception.

1. Is the title eye-catching?
2. Does the content have the potential for secondary dissemination?
3. Is the content practical?
4. other

Menu/Message Analysis

This is to provide a basis for analyzing behavioral events in which users actively interact with official accounts.

1. By comparing the daily menu bar clicks, we can understand the user activity and evaluate whether the menu bar column configuration maximizes the user's enthusiasm for interaction.
2. Many activities will provide activity entrances through the "keyword reply" method. By analyzing the number of people who send messages, you can understand the user's participation in the activities.
3. Message keywords , regularly understand the sending frequency ratio of "custom keywords" to analyze the user's participation in daily activities; regularly understand the sending frequency ratio of "non-custom keywords" to analyze the user's potential interaction demands, so as to gain insight into the user.

Interface Analysis

When a public account authorizes a third-party plug-in platform, this dimension of data can be used to understand the problems with the interface calls and provide timely feedback to technical colleagues for resolution.

Final Thoughts

If you create a public account to build a personal brand, then your core competitiveness must be the accumulation of your own knowledge system . If your internal "input" is not enough, your "output" will eventually run out. People who have content and knowledge accumulation will not miss the so-called bonus period. (Of course, even if you don’t have a WeChat public account, you can still shine on other platforms, such as Zhihu and Fantuan. Don’t limit yourself to WeChat.)

And if the public account you create is the official media of a company, think about it carefully, for most enterprises and institutions, are there more comprehensive online touchpoints than public accounts? Cashless payment for food, accommodation, travel, entertainment and shopping, various government affairs inquiries and processing... The comprehensive convenience of WeChat cannot be replaced by anything else, at least in the next three years.

SO, don’t use the general environment as an excuse for why you don’t do well. Nothing will always move upwards, and spiral development is inevitable.

There are still many ideas for in-depth discussion on data collection and analysis in this article today. If you are good at WeChat data analysis, you are welcome to communicate with me!