Ali 10' s pension operation data-based activities are that simple?

How to use data to evaluate activities? The author of this article will enjoy ~ 1 by combining his own experience and your analysis. Background E-commerce platform A, established nearly 10 years ago, is not the top in the industry, but it is also a bit prominent in the segmentation field. This year is not very good. The daily activity of platform APP has dropped from 50. 1 10,000 at the beginning of the year. Since its operation, it has been dragging 400,000 underwear, and the boss has racked his brains to coax investors. Meeting "This is no good, we have to do something." The boss stubbed out half the unburned cigarettes. "Talk about several cooperation platforms and exchange resources for traffic?" "throw a few content KOL or throw a few advertisers?" "Do a few more promotions, and cooperate with Push to be more daily." BD, marketing and operation put forward suggestions, and the boss was silent. 18, the mobile terminal is depressed and the industry is saving. No one dares to invest a lot of gold and platinum. Promotion may stop loss, but to completely solve the problem, new traffic must enter. After a long silence, the boss asked, "How many old customers have lost their orders under the platform these years?" "89 million." "Find a way to get these users back." 2. Boss Fa is right. From the experience, recalling lost old customers must be cheaper, more accurate and have better traffic quality than pulling new ones. 1. With what? 89 million old customers left their mobile phone numbers, names, addresses and other information during the registration and ordering process. It is not difficult to contact them. The simplest, efficient and low-cost way should be SMS. Tell them what? What kind of text messages can attract them back? As far as e-commerce is concerned, there are generally three ways to play: promotional offers, such as: 30% discount on summer explosions, and SK-2 fairy water is only sold? 698, hot speed, slow speed! ; Send coupons, such as: When the weekend comes, I will give you a 88 yuan Gourmet coupon, and click the link to get it! Receiving goods, such as: you have been here for too long, and the boss said I was negligent. I'll give you a ray mask worth 59 yuan. Come and get it! The first method has low cost (only SMS fee), but poor response. It is generally used to cooperate with promotional activities and is not suitable for regular customers to recall. The second and third methods generally respond well in SMS or push. However, as a non-proprietary platform, coupons mean a high subsidy cost of 1: 1. At the same time, for the lost old customers with uncertain purchase demand, coupons are not just needed and have limited appeal. Relatively speaking, "free explosives" is not only common, but also the cost of large-scale procurement is lower. So we use the third method: SMS to inform these lost customers to come back for free gifts. 3. What attracts them? Where can I get it? APP？ The lost customers are all users who haven't visited for 180 days, and most of them have uninstalled the APP. The cost of reactivating the APP is too high and unrealistic. Don't go to the APP, where to go? In the e-commerce industry, "free collar" activities are generally accompanied by social games such as group fighting, sharing and bargaining. And further buffer costs through fission and innovation, such as Pinduoduo's help in bargaining and 0 yuan's daily delicious fruits. The best hotbed of social fission is undoubtedly the applet. As an extension of the WeChat ecosystem, the A platform applet is lightweight and complete, and it is the first stronghold to undertake the return of old customers. To sum up, we will inform the lost old customers to return to courtesy through SMS, introduce small programs, and generate fission through social play. The activity flow is as follows: In the above-mentioned flow, starting a tour group or joining a tour group does not mean the end of its formation or failure, whether in the main state or the guest state. When users participate in this stage, we have obtained the permission of WeChat OpenID and pushing template messages. At this time, you can push promotional products, coupons or other activities according to the user's status, promote them, and undertake further transformation of users. The background and planning of the incident are very clear. Let's talk about how to use data to track and evaluate the benefits of events. 3. Data evaluation of the above-mentioned activities is carried out from four dimensions: response data fission data value tracking cost evaluation. These four dimensions are also basically suitable for evaluating all micro-operation activities. Let's expand one by one according to the previous activities and give examples. In view of business rules, the data will be trimmed within a reasonable range. 1. Response data Response data is what we often call a "funnel". Suppose we put an advertisement in today's headlines. Within a period of time, 1000 people clicked on the advertisement, 200 people registered after clicking, 50 people downloaded the APP according to the prompts after registering, and finally only 10 people completed the consumption conversion in the APP. From 1000 to 10, it is a simple funnel. Any activity, from the first specific action of the user to the final transformation after it is achieved, is a funnel. Funnel shows the direct effect of an activity, and decomposes the effect into different steps, providing each level of attribution for the result. The following is the transformation funnel of this old customer recall activity: the "password input ratio" is low, which may be because the short message copy is not sharp enough and the product selection is not attractive enough; The ratio of "number of people sharing links" to "number of people opening groups" is not bad, which shows that after the user enters the password, the guidelines and procedures for forwarding the group opening are clear and concise, and the user's operating threshold is low; The "aggregation rate" is 7.79%, which is lower than the industry level. Guess it may be because there are so many people in a group of three that it is too difficult, or it may be because these users are troublesome or unwilling to abuse their social value. Through the funnel, we can not only see the overall transformation level of the activity, but also study the data performance of each link in detail, and see at a glance where the problem lies. "Where is the problem" is only the first problem solved by the funnel, and more importantly, "how to solve it". The above is the "trunk" of the funnel, and the following is the "branch" of the funnel: for the problem of "low input password", we guess it may be related to copywriting, product selection, sending time and so on. Taking the choice of products as an example, in order to compare and test, we set a small number of people (5000 people) for each group as the control group before the activity, and pushed short messages to verify what kind of products are more attractive to this group. Obviously, users under the age of 24 are more willing to get eye drops for free, and masks are more attractive to users aged 25-29 and 30-39. More than 40% users prefer toothpaste. We found that the above table not only contains the variable "product choice", but also contains the intersection of the two variables "age stratification" and "product choice". In an activity, the variables that affect the outcome of a link are often numerous and intertwined. Taking this activity as an example, the factors that affect the password input ratio are not so much a single variable as the combination of variables such as age and product selection, age and copy, and even the optimization of all related variables such as age, product selection, copy and sending time. Therefore, the "branch" of the funnel is to seek the optimal combination of various variables. In practice, the combination of variables will produce exponential growth, and 35 variables may combine hundreds of situations, so it is unrealistic to test them one by one. Therefore, variable optimization must be based on business judgment. For example, in this case, from the perspective of business experience, users of different ages should have different needs for goods, so we made a variable test of "age" and "product choice"; Similarly, we think that there is no clear correlation between "product selection" and "SMS sending time", so we take "SMS sending time" as a separate test variable. Whether it is the verification of conjecture or the mining of insight, data must be just a tool of business, not the business itself. 2. Fission data In the planning of an operational activity, if fission is not the starting point of the whole activity, it should at least be regarded as a key value link. Whether it's red envelope sharing when you are hungry, Pinduoduo-style help bargaining, delicious fruit in 0 yuan every day, Ctrip's help to grab tickets, recent phenomenal psychological test of Netease in friends circle, My Coffee Shop with coffee, these textbook-level activities or games, regardless of transformation orientation or communication orientation, are all around the value point of "socialization". Even for activities outside the WeChat ecosystem, such as Alipay's Spring Festival gathering in Five Blessingg, the highly social "sending blessings" is the core source of its participation in the transformation. The reason for launching the activity is that there are four or two opportunities to rely on social communication behaviors such as forwarding, sharing and praise. Having said that, I just want to demonstrate that the tracking and analysis of "fission data" is by no means dispensable at the moment when operational activities are becoming more and more socialized, and it has gradually become an important data indicator for evaluating the effect of activities. Let's go back to our case. As can be seen from the activity SOP diagram, there are two main fission modes in this activity: main fission: users will divide the activity into three different groups in order to get goods for 0 yuan. When group members see the link, they can also share three opportunities to get 0 yuan to open the group and become the new main state. Guest fission: after the main user starts the group, new users join the group. If successful, a guest fission is completed. The first fission shows the first fission effect brought by SMS users, and the subsequent fission refers to the uncontrollable fission caused by the active link being pushed to 39859 users by SMS. Because of the buried point of the active page, the subsequent fission cannot be traced layer by layer, although it is necessary (the groups forwarded by users of different ages are likely to be completely different in quality). The number of shared and open clusters in the subsequent fission represents the main fission; The number of participants represents the fission of guests. As can be seen from the table, the first round+subsequent fission * * * counted 39859+ 13477=53336 groups, but only1121people clicked on the link, with an average of 0.2/in each group. Compared with the first round of fission, the per capita sharing times, the proportion of participating groups and the clustering rate of subsequent fission have greatly decreased. On the one hand, the information contacted by the two groups of users is different, on the other hand, the users who received the SMS are old customers of Platform A, which is more accurate, while the users generated by fission are uneven, disoriented and poor in accuracy. We also analyzed the problems of low composition rate and small number of participants in the group. To sum up, the fission of this activity is not ideal, and the display, commodity selection and copy of sharing links need to be optimized. If the optimization is still unsatisfactory, the logic of the whole activity may have to be reconsidered. 3. With the response and fission data of value tracking, the operation walked into the boss's office with confidence. He briefly reported the main points in the transformation and fission data to his boss, made business attribution, and put forward some suggestions and arrangements for the optimization of the next implementation plan. The boss looked at the data and nodded slightly, thinking about something. At this time, the small operation made a summary and angered the boss. "I think the effect of this activity is ok. If the logic of fission is optimized, it can be carried out in large quantities. " In fact, we should not draw any conclusions before the value tracking of converted users is completed. The so-called "transformation" is not only the completion of group formation, but also the loss of users in any link such as sharing, registration and joining the group. As long as it has something to do with us, it is considered as transformation, but the degree of transformation is different. The transformation of each user means that the costs and subsidies of real money must be filled with the profits brought by subsequent purchases. If the user repurchase and activity brought by the activity are extremely poor, then no matter how high the conversion rate is, the activity is a failure. Because your activities attracted a group of people who were scared by the operation industry-the wool party. In order to improve the subsequent performance of users after conversion, operators usually use push or SMS to further encourage them. For example, in this example, we will personalize coupons, preferential products, activity links, etc. for users who have successfully traveled, failed to travel, successfully participated in group tours, failed to participate in group tours, and shared links less than three times. Based on the conversion degree within 24 hours according to their behavior tendency, to promote the subsequent conversion and try to take over these users. The data of value tracking is complex, which can form a huge value tracking system based on user stratification, time stratification and behavior stratification. Considering the space, the author has simplified the value tracking of this example, which is divided into the following two parts: (1) Key value tracking This table has three main points: 1) Who is the new customer? However, it must be traced layer by layer. In addition to the age stratification of SMS push, it also needs to include the stratification of conversion degree, so that the value difference between users with different conversion degrees can be tracked. As shown in the above table, we can see that the repurchase frequency of group users is higher than that of group users; The value of fission users is not as good as pushing people. 2) How long to pursue depends on the frequency of product use and the life cycle of users. General e-commerce tracking for 90 days or 180 days. If conclusion data is urgently needed for activity optimization, short-term tracking data can be provided. 3) Pursuing what must be the most critical value indicator for tracking users to products. For self-media, it may be clicking, forwarding, and rewarding; May be active and pay for the game; For the e-commerce platform, it is naturally transaction data such as GMV, order number and customer unit price. So far, we have obtained how much GMV each layered user contributed in different time periods, and multiplied it by the transaction rate of the platform, so it is easy to calculate the profit obtained by the platform. Through these profits, on the one hand, we can intuitively judge the value of different layered users, on the other hand, we can judge the income of the activity by comparing with the cost, and decide whether the activity can be productized or copied on a large scale. In the case of simple products and small activities, the above value tracking can basically prove the value of activities. However, if the product ecology is complex and the user's consumption behavior is diverse, it is necessary to track the user's behavior in layers. (2) Behavior Hierarchical Tracking In the same product, different users have contributed different values to the platform based on different behaviors, and * * * has jointly built a complete ecosystem. We take Zhihu, the most representative UGC content platform, as an example. Its user behavior and corresponding value can be roughly divided into: compared with Zhihu, the ecology of e-commerce platform may be more complicated: in the e-commerce ecology, it is not just as simple as "buying" and "selling". Search and buy users have contributed most of GMV, which is the foundation of the platform's survival; Users who like to "stroll" are keen on discovering shopping. In addition to contributing considerable GMV, they also feedback the popularity of banner, modules and products with different themes through click-and-stay, thus pointing out the direction for operation. Those users who are keen to fight for the group and help bargain gather on WeChat, exchange the benefits of the platform by spreading labor, and constantly introduce new traffic to the platform; Even at present, most e-commerce platforms have introduced content sub-ecology, such as strategy and evaluation, and users who like to write evaluation results are the core support of this sub-ecology. Therefore, we can't judge the value of users only by purchasing power, that is to say, we can't think that the activity failed because the users attracted by the activity didn't buy enough. Because these users are likely to contribute other important ecological values. From the data point of view, the users brought by this activity are more prominent in opening a group, helping to bargain, and paying attention to the service number. It shows that users gather in WeChat applet, which has good communication value. It is speculated that the activity game is a team battle, so the traffic is more or less wool, which also reduces our expectation of its repurchase value. These users are more suitable for peripheral communication for the platform in small programs. 4. Cost data "Cost data" and "value tracking" are two aspects of the job. The difference between value and cost basically represents the initial income of the whole activity. As far as this activity is concerned, the cost mainly occurs in three places: SMS data, the loss of old customers touched by SMS is about 0.03 yuan/article; Free commodity cost subsidy, all group orders, the commodity cost to be paid is around 20 yuan; Follow-up incentive subsidies; After the user is converted, we may push some coupons to promote repurchase, and coupons are the main cost in this process. The third step is to track the total GMV of the repurchase, get the profit through the platform rate, and then subtract the sum of the above three costs to get the activity profit. The logic of cost data is simple, and the only thing worth noting is whether it should be stratified. Different layers have different user values and different costs. If the same layer is used to track the value and cost, the ROI of different layers can be calculated, the break-even point can be found, and the profit of the whole activity can be improved through selective push. However, in many cases, it is difficult to achieve hierarchical ROI optimization based on limited user list and data embedding points. Therefore, whether it is stratified depends on the business situation. Finally, show a mind map: