Data analysis is one of the core competencies of operation, especially in activity operation. The application of data analysis can help the operation of activities from subjective to objective, from chaos to controllability, from defects to perfection, and it is a compass that runs through the activities and guides the operation of activities. Today, I will systematically talk about how to make good use of data in the operation of activities, involving specific methods of data analysis, data analysis of different types of activities, cost-benefit evaluation of activities, and AB comparative experiments. I hope it helps you. The full text is long, and the structure is as follows: 1. Basic skills of operational data analysis When it comes to activity operation, it is easy to think of creativity and crazy play, and behind creativity and planning operation, data analysis is also needed to provide strong support. Many activities can't be said to lack the ability of data analysis, because ta has no awareness of data analysis. The consciousness of data analysis refers to paying attention to data, attaching importance to data, and making decisions and actions based on data analysis in all aspects of operation. Paying attention to operational objectives, reducing experience dependence, increasing data support, and first establishing the awareness of data analysis are the prerequisites for data analysis. With the awareness of data analysis, let's talk about data analysis ability. Data analysis ability pays more attention to analysis methods: how to disassemble positioning problems? How to find the analysis dimension? Common analytical method model? This is the content that should be focused on when learning to apply data analysis. In addition, when it comes to data analysis ability, many people will think of data analysis tools. For operations, the importance of tools will be much lower. Excel is basically enough, and other tools have extra points, but they are not necessary. This paper focuses on the basic skills of data analysis: analysis process and analysis method. In the process of data analysis, the end point of data analysis is to draw the conclusion of the reasons and form suggestions/decisions to guide the subsequent direction and actions. There are three steps to reach the end of data analysis: the first step is to clarify the problem or goal. This is the starting point of data analysis, which determines the center and direction of data analysis and is the premise of effective data analysis. When defining a problem, we need to avoid the preconceived definition of the problem, and make clear the scope and goal of the problem from the phenomena and data, which can guide the subsequent analysis scope and thinking. The second step is to disassemble and analyze the reasons. Based on the identified problems or objectives, further decomposition analysis is carried out to locate the key factors that cause the problems/affect the objectives. In most cases, there will be multiple influencing factors, so it is necessary to judge and analyze the influence size and verify the role of the factors. Efficient disassembly analysis needs to combine data analysis methods with practical experience. The methods are shared as follows, and the experience is gradually accumulated. The third step is to draw a conclusion. The value of the last step of data analysis is to find the solution to the problem, the suggested method or the path to achieve the goal, and to output conclusions and suggestions. Conclusion refers to the key factors of the problem, the influence mode/size of the factors and other related factors, and suggestions refer to how to influence the key factors, what actions need to be taken, and the ideas and strategies of action. After data analysis, we should continue to act. Data analysis methods There are many models of data analysis methods, which need to be selected flexibly according to the analysis requirements. For complex operational research analysis problems, it is generally necessary to adopt a variety of analysis methods and models. Here are three commonly used essential analysis methods to share with you. Method 1: Comparative distribution analysis method Comparative distribution analysis method is the most commonly used analysis method, which is helpful to find the problem of definition difference in the early stage of analysis. The core of comparison is to solve the problem of how to compare effectively and find out the clear characteristics of differences. Compared with the outside, there are more differences, while the internal comparison focuses on change. Specific comparison can be compared by scale: focus on total/average/median, fluctuation: focus on variance/standard deviation/extreme range, trend: focus on chain/year-on-year/change. Scheme 2: Path Funnel Analysis Path Funnel Analysis is especially suitable for activity operation, which can effectively monitor the process effect and locate key factors. Using the path funnel analysis method, we first need to split the activity stage, the propaganda arrival stage, the participation and sharing stage and the fission transformation stage, and then make clear the user behavior of each stage, especially the high-value user behavior, and then sort out the corresponding data indicators to form a complete path funnel. In actual monitoring, combined with comparative analysis, pay attention to variation differences and abnormal data, analyze and apply them. Method 3: dimensional disassembly analysis plays an important role in locating the causes of specific problems and the components of disassembly targets. Many times, we are faced with the appearance or result of the problem, and data analysis needs to find specific reasons. At this time, it is an effective means to find the reason by disassembling the overall data and problems. Common split dimensions, such as time dimension, channel dimension, user stratification, etc. For example, the decline in daily sales of promotional activities can be disassembled from the channel dimension, whether it is the decline of WeChat or APP, or from the order dimension, whether it is the decrease in order volume or the decrease in order amount. Through continuous disassembly, the final cause can be located and the subsequent operation can be guided. Data analysis is a continuous work, which needs attention. At the same time, data analysis is also a complicated work, which requires learning methods to accumulate experience and lay a good foundation for data analysis. Second, the data analysis that runs through the activity The application of data analysis in the activity operation runs through, and every link can play the value of data. How to use the data before the activity? What do you need to do before the activity? First of all, the activity is established, and it is decided to do an activity under a specific goal and background; Then define specific objectives and resources to guide activity planning and resource input; Then it is to plan and communicate activities, form an activity plan and promote implementation. What role can data play in it? The first is the early stage of the project, which needs data support to explain the necessity. Why do you want to do this activity? What problems can be solved? Through investigation and analysis, data disassembly and operational insight, the necessity of explaining activities can be supported. Secondly, in the stage of determining goals, data is needed to support clear guidance. How to set activity goals? Is it reasonable? Through the data of competing activities and historical activities, and by comparing the differences, the overall goal is defined, the activity process is dismantled, the resource input is determined, and the effects of each link are gradually estimated to improve the accuracy of the overall goal. Finally, in the activity planning stage, data support is needed to improve the feasibility. Why do you want to design activity flow rules like this? How to ensure the effect of the activity? From the perspectives of time rhythm, scene channel, activity reward, user characteristics, process characteristics, cost and benefit, combined with data comparison/disassembly, it supports the formulation of activity play process. How to use the data in the activity after the preliminary preparation and implementation, and then the online operation, at this time, activity data monitoring and iterative optimization become the focus. Activity data monitoring ensures the control of the functional state and operation effect of the activity and the stable operation of the activity. Activity monitoring pays attention to the process indicators and ultimate goals of each link by disassembling the activity process, and can also be further divided from the time dimension, the channel dimension and the user dimension on the basis of the overall data to monitor the activity effect more accurately. On the basis of data monitoring, it is necessary to adjust and optimize the activities in combination with data analysis. There are many angles to find optimization ideas, such as looking at data fluctuation in time dimension, paying attention to abnormal data promotion or decline, and analyzing specific reasons; Look at the feature differences from the user/channel dimension, pay attention to the activity data differences of different channels or different types of users, and focus on the dominant channels and users; We can also look at the improvement space from the historical data dimension, and make clear the effect and improvement space of current activities and determine the value and investment of optimization according to the historical activities and industry competing data. How to use the data after the activity? After the activity, data is an effective way to reflect the effect of the activity and make a summary. The results of activities should highlight the results of core indicators, auxiliary indicators and process indicators, compare the differences between goals and reality, and at the same time disassemble the dimensions to locate the specific reasons for the good/bad results of activities. Is a certain channel bad? Or is the participation effect of some users more than expected? Only in this way can we gain valuable experience and guide the planning and operation of more follow-up activities. Data analysis plays an important role before, during and after the activity, which can be said to be related to the success or failure of the activity. Pay attention to data and make good use of it. Third, the cost-benefit of returning to the target can not be separated from "money" in the operation of the activity, including both the cost and expenses of the activity, as well as various benefits generated by the activity, and the relative relationship between the cost and the benefit, that is, the input-output ratio (ROI). First talk about the cost pull-up/conversion/promotion activities in the activities, which mostly involve the cost and expense input, and need to be accurately calculated and evaluated: (1) The pull-up activities have the cost of acquiring new customers, that is, how much it costs to bring a new user, also called customeraccessioncost(CAC). The total cost of the pull-up activities (including promotion cost and prize cost) is obtained by dividing the number of new users brought by the activities. CAC reflects the cost of acquiring users, which affects the sustainability of pull-up activities. What really determines the effect of the promotion activity is the relative relationship between user acquisition cost and user life cycle value. It is also easy to understand. The pull-up activity can cost 100 yuan to acquire a new user, and the end user contributes 1000 yuan in product consumption, so such pull-up activity must be done vigorously. (2) The conversion activity has the first conversion cost, that is, how much the new user needs to complete the first order. The first conversion cost reflects the effectiveness of conversion activities and is also a way to judge the quality of users. The higher the subsidy cost of user conversion, the worse the quality of users, and the expected follow-up retention effect is not good. (3) Promotional activities have promotional costs, that is, how much it costs to bring a certain amount of sales, such as subsidies 1 10,000 and sales of 1 10,000, which is the cost of promotional activities. We can also pay attention to the cost of each promotion order/user from the order dimension and the user dimension, so as to monitor and analyze the cost of promotion activities more accurately. In the long-term operation of product and user life cycle, there is a very important concept of revenue and ROI: user life cycle net value: user life cycle net value = user contribution value _ product cost. The value contributed by users, that is, the amount spent by users on products and the profits they bring, can also include friends and word of mouth in a broad sense, while the cost of products includes the aforementioned innovation cost, transformation cost and promotion cost. When a user uses a product for the first time, the net life cycle value of the user is generally negative. As users actively consume and contribute more value, users' net life cycle value will become positive. The larger the users with positive net worth, the higher the product value. In specific activities, it pays more attention to the ROI (input-output ratio) of activities, that is, income divided by cost. (1) Pull-up Activity ROI In the pull-up activity, the customer acquisition cost needs to be invested, but the quality and life cycle value of the acquired users cannot be directly measured. It is necessary to refer to the user's performance and life cycle value in the product to calculate the ROI of the pull-up activity. For example, the average contribution value of new users brought by the previous promotion activities is 100 yuan, the customer acquisition cost of this activity is 1 10 yuan, and the ROI of the activity is less than 1, indicating that it is a loss and this activity cannot be continued. (2) Transformation Activity ROI transformation activity is aimed at new users and promotes new users to complete the first single transformation. Although it can bring sales, there are many large subsidies for new users, and it is difficult for the ROI of conversion activities to be directly greater than 1. Transformation activities need to focus on short-term return on investment and long-term return on investment. Short-term ROI refers to the sales of users in conversion activities divided by the cost of conversion activities, while long-term ROI focuses on the contribution value of users in a certain period, and then removes the cost of conversion activities. Generally speaking, long-term ROI is more important. (3) Promotion activities ROI Promotion activities are the main forms of activities to enhance sales and user value, and pay more attention to short-term sales revenue and ROI of activities. In addition, there will be new user conversion and silent user recall in the promotion activities, and these users can pay due attention to the subsequent retention and long-term benefits. Doing activities means spending money, paying attention to cost means "spending money clearly", and paying attention to income and ROI means "spending money effectively" Cost-benefit thinking is necessary and should be paid attention to in operation. Fourth, the activity experiment of iterative upgrade understands A/B testing. A/B testing is a very important tool for data application-driven operations. A/B test speaks with facts, which is the thinking of comparative analysis. Scheme a and scheme b are the best. Strictly speaking, A/B testing refers to the random use of multiple schemes for the target population with similar composition and characteristics in the same time period, and the final evaluation and judgment of the best scheme through the comparative analysis of the results of the schemes, and then the formal and comprehensive use. In summary, the premise is uncertainty, the core is univariate, and the conclusion is survival of the fittest. A/B testing can effectively support the design, operation and iterative optimization of operational activities, especially long-term or periodic activities. With the help of A/B testing, the effect of activities can be effectively improved. From the design of activity flow and rules, to the selection of activity rewards and games, to the determination of activity page style and copy, A/B testing can be used to verify and optimize everything. Application of A/B test The application of A/B test can be divided into four steps: step 1: define the goal and form the hypothesis. To do the A/B test experiment, we must first make clear what to verify, and have pre-analysis and judgment, instead of blindly taking out multiple schemes for testing. After all, A/B testing also needs a certain cost and time period, so try to choose the best, not one in a million. Step 2: Determine the indicators, select users, determine the key elements that need to be tested and verified, and at the same time clarify the key indicators that affect, so as to effectively compare the test conclusions after the subsequent tests are completed. At the same time, A/B testing is aimed at the verification of uncertain schemes, so as to avoid large-scale users as much as possible, and some users can be randomly selected for A/B testing. Step 3: Design the scheme. The most important thing of online experiment A/B test scheme is to ensure the uniqueness of test variables. Except for the variable elements to be verified, all other schemes are the same, so as to avoid interfering with the analysis and judgment of the test results. Step 4: Analyze the results. After the A/B test experiment has the result data, it is to analyze the advantages and disadvantages of each scheme and decide the best scheme, mainly by comparing the key indicators initially determined, and also paying attention to the process indicators and experience-related indicators. The data analysis of activity operation ends here. The basic skills of data analysis should be consolidated, the data analysis in the process should be thorough, and more attention should be paid to cost and benefit, making good use of data experiments to improve the effect. I don't need to say more about the importance of data analysis in operation. In your daily work, you will get better feedback if you apply data consciously and methodically. Author | Wu Shang Changhai Pirate Ship Owner