Are activity operations still following hot topics?

Internet companies have transplanted the form of traditional offline activities . During festivals, they will launch popular activities to boost user activity and product sales. So, are your current activities still following the holidays? In fact, the pace of data-driven operations is quietly accelerating.

During the just-concluded "National Day" holiday, businesses from all walks of life saw this rare "Golden Week of Consumption" and did their best to launch various promotional activities. As the digitalization process of enterprises continues to accelerate, the operational activity rhythm system, which has undergone many changes, has ushered in an accelerated upgrade driven by data.

Event operations account for a very large proportion in the operations of various industries, and regulating the frequency and density of activities has become a thorny issue. On the one hand, we hope that the more activities there are, the better the performance will be. On the other hand, too many activities will lead to a decrease in users’ “sensitivity” and “response rate” to the activities.

In the course of the development of the Internet, event operations have shown four generations of distinctive event rhythm systems.

Holiday Type

Internet companies have transplanted the form of traditional offline activities, and will launch popular activities during festivals to boost user activity and product sales, such as Double Eleven discounts, 10% extra phone bills for recharges, membership days, etc.

The most popular festivals include:

• Traditional festivals such as New Year's Day, Spring Festival, Dragon Boat Festival, National Day, Christmas , etc.
• Valentine's Day , Singles' Day, Chinese Valentine's Day and other themed festivals;
• Some companies have created their own anniversary celebrations and monthly membership days.

Macro

When running festival-type events, the effects of each event vary from time to time. Some conclusions can be drawn from the data analysis of the events.

For example: when the total number of remaining days of (video, bicycle) membership of all users is large, the effect of the promotion will be worse, and when the total number of membership days of all users is small, the effect of the same promotion will be better.

Therefore, it is natural to develop a set of activity strategies based on macroeconomic and ecological monitoring data:

• Carry out activities at the "optimal" time node.
• The activity materials are prepared in normal times, and once the indicators break through the threshold, the activity can be launched at any time by configuring the keywords .

Similarly, two experiences were summarized in practice:

When a data curve (indicator) is not appropriate, it must be optimized and adjusted. For example, when doing a bill installment activity, it goes through multiple adjustments, from the total bills of all users, to the total bills of users with a record of bill installments, and finally to the total bills of a small group of users who monitor 50,000 bill users.

Even if the activity is carried out at the "optimal" node, the value trend of each user will be different. Often, when most users are present, some users happen to be at the "worst" time for their own parameters. Therefore, if we can customize marketing strategies for each user, we will be able to further tap into the value of users.

Microscopic

The most basic work in researching and operating the unique attributes and preferences of each user is "labeling". In addition to conventional demographic attributes (age, gender, province and city) and interest preferences ( games, videos and products), more behavioral attributes should be collected, such as game win rate, number of ad skips, average account balance, etc. Based on the established label system, the operation team can launch personal exclusive activities similar to “Thousands of Faces for Thousands of People”.

During the implementation process, it is necessary to regularly query changes in personal tags and their attribute values. Once the marketing conditions are triggered, the preset activities can be pushed.

For example, if a scheduled task is performed in the early morning of a certain day and the number of credit card transactions of user A in the past week is found to be less than 10% of the 15-week average, a small amount of limited-time credit card reward will be given to the user, which will help increase the user's card swiping activity.

Real-time

Retrieving user attributes is usually time-consuming and is usually performed in the early morning. With the development of real-time computing platforms in recent years, it has become possible to push marketing activities , especially promotional activities, in real time.

For example: if you search for an unfamiliar product (perfume, fishing rod, grinding machine, etc.) that you have never searched for before on the Taobao homepage and click to view it, Taobao will respond to you in a very short time: recommended products in this unfamiliar category will be displayed on the homepage and in the recommended positions, sometimes even the product you just clicked on. The shortest sampling time for the whole process is only 17 seconds.

Similarly, you can also launch store discount coupon activities.

Source: Analysys