With big data, machine learning and various data analysis algorithms flying everywhere, many people are lost in the ocean of various algorithm implementation, various operation means and various anxiety. Today, I would like to talk about my understanding of three things in the topic.
1. What is operation
Operation is responsible for creating short-term user value + assisting products to improve long-term value
1.1 Classification by operation type
1.1.1 Content operation
Build a virtuous circle around the production and consumption of content, and continuously improve all kinds of data related to content, such as content sorting, content browsing, content interaction, content communication
1.1.2 User Operations
It establishes a virtuous circle around the new – retention – active – spread of users and the value supply relationship between users, and continuously improves all kinds of user-related data, such as number of users, active users, elite users, and user retention time, etc
1.1.3 Activity operation
Plan one or a series of activities, resource confirmation, publicity and promotion, effect evaluation and a series of processes to promote the whole process of the project, schedule management and implementation
1.1.4 Product operation
Through a series of various operational means (such as event planning, internal and external resource expansion and docking, product plan optimization, content organization, etc.), the specific data of a product can be promoted, such as installation volume, registration volume, user access depth, user access frequency, user relationship to data, Posting volume, etc
1.1.5 Other operations
For example: new media operation, APP store promotion operation, SEO/SEM operation, advertising operation/traffic operation, Taobao store operation, QQ group, group operation, etc
1.2 Classification by operation content
1.2.1 Tool Products
Efficiency and experience are emphasized, and the product is usually more important than the operation. Therefore, in a long period of time, the most important focus of the operation side is the user growth. The main means include channel promotion, BD and some activities, etc., and the operation side will deal with more data
1.2.2 Social/community products
Social/community products focus on social atmosphere, topics and gameplay, and are a product form that requires equal emphasis on operation and product
1.2.3 Content products
Continue to create unique, high-quality content, and package it in a way that makes it easier for users to consume
1.2.4 E-commerce products
About the operation of commodities and categories, we should pay attention to the problems including what kind of commodities to choose for sale, which commodities to focus on in the selling process, how to formulate the pricing strategy of commodities, inventory management and supply chain management of commodities, etc. Planning and implementation of various promotional activities; Promotion and flow construction; Customer care and customer retention.
1.2.5 Platform Products
Focus on strategy and customer retention. For example, Taobao merchants may need to be divided into many categories according to different dimensions such as regions, goods sold, and customer unit price, and then maintain them respectively.
1.2.6 Game products
Promotion, all kinds of docking channels, all kinds of conversion rate, all kinds of staring data; Revenue, for example, for a group of people with a higher probability of paying in a game, the game company may have a dedicated team around this group of people, through various strategies and operations to promote this group of people to pay.
1.3 Classification by operation work content
1.3.1 User Login, Traffic diversion, and Conversion
1.3.2 User Maintenance
2. Value of data to operations
- Data can objectively project the current status and stage of a product
- If you get something done and it doesn’t work well, the data can tell you, what’s wrong with you
- If you want to achieve a goal, data can help you figure out the best way to get there
- Extremely detailed data analysis helps you understand your users and have more control over the ecology of your site by breaking down layers
- There may be clues and eggs hidden in the data to make something better
3. What is data analysis
3.1 Basic Ideas
- Comparison: You can compare users, services, and time points
- Ratio = target resources/available resources
- The difference between the two: comparison can only be the analysis of two data with the same caliber and dimension, such as amount to amount ratio, number of people to number ratio, duration to duration ratio, etc. However, the ratio can be directly analyzed by cross-dimensional data under the same caliber, such as amount to number of people, amount to time, etc. For example, when analyzing operating costs per user, the amount of money spent is the target resource, and the number of users driven by that money is the available resource
3.2 Work Content
3.2.1 Data Planning
Metrics need to be defined according to the objectives of the operation. Indicators are used to measure specific operational effects, such as UV,DAU, sales amount, conversion rate, etc. Indicators are selected from specific business needs, events are summarized from the needs, and indicators are corresponding to events. Dimensions are familiarity used to break down metrics such as AD source, browser type, region visited, and so on. The principle for choosing dimensions is to document those dimensions that are likely to have an impact on the metric
3.2.2 Data collection
- Acquisition scheme: buried point, no buried point, visualization buried point
- Data source: external data (Excel, monitoring platform, data of other websites), internal data (system logs, database data)
3.2.3 Data analysis
A. The number of visits, the number of clicks is to have A macro understanding of the overall situation of the product; B. Conversion rate. Funnel model transformation artificially sets a specific analysis target — transformation path — according to business objectives and business core processes. Funnel model is a visualization method for transformation c. User portrait, user group user portrait, is to visually present the appearance of users, the appearance here refers to the characteristics of users, for example, geographically concentrated in second-tier cities, the age of users around 20-30. When we do the user portrait, the user in our mind begins to appear differentiation, according to the different dimensions of the whole user group after the segmentation, each user group has its own different portrait.
3.2.4 Data visualization
A. Bar chart: quantity relation, quantity ratio b. Pie chart: total score relation, partial and total comparison, proportion C. D. Scatter plot: The distribution of a variable in two dimensions E. Stack diagram: see the data of several categories and the proportion in the population, and see the comparison of the total situation after the data of several categories are added together
3.2.4 Data automation
The first four steps can be automated to provide operational staff with continuous reports of data analysis
3. python
This is a universal language, the specific introduction of online Baidu, here just say about the content related to data operation. A. crawler: when you need to get the competition site data, through the crawler can easily access to the other side of the data, analysis processing b. web, mobile automation: when you need deal with repeated page or phone operation time, can use this code instead of manual operation, improve the operating efficiency of c. qq, WeChat robot: When you operate a QQ group or wechat group, you can write robots to complete some simple replies or automatic operations, such as regular message sending and group message sending. D. Data analysis: Panda package can carry out very convenient data operations on data, and only a few lines of code can process very complex data contents. Machine learning: The SciKit-Learn package encapsulates a number of machine learning model algorithms that can be implemented using F. Visualization: Matplotlib contains rich graphics
Example 4.
Now that you’ve had a first look at data, operations, and Python, let’s walk through the process with an example, the implementation of which will be explained in a later article. Objectives of operation: identify the most valuable users among customers and carry out refined operation for this user group A. Index: consumption frequency, last purchase time, and last purchase amount B. Data collection: Obtain the above three data from the database C. Analyze by RFM model D. process #5 by K-means clustering method of Python. To sum up, if the operation is compared to a war, data is the intelligence data in the war, python is the tool to decipher the data of the intelligence, and a step faster in the war is the victory.
Follow public account
