How does big data drive products and operations?

Maybe everyone has had this experience: after you have logged into an app or browsed a product, you may often receive promotional or activity text messages about this app or this product on your mobile phone. So how is this kind of reminder function implemented? Let me share my experience here.

Everyone knows that all reminder behaviors are for attracting new users , retaining existing users , promoting activation and conversion . To achieve this goal, we not only need to quickly notify target customers, but also evaluate the effectiveness of the reminder.

Based on the above requirements, we first make a mind map (combined with the situation of our company), as shown below:

First of all, the reason for posting (why) is to attract new customers, retain customers, promote activation and conversion. I will not discuss this in detail here.

Secondly, to whom it is sent (who), from the above picture we can conclude that: to screen out our target customers, we need to know the login habits, consumption habits and behavioral preferences of each member over a period of time. The user portraits and user behavior records under big data record and analyze each customer's login, consumption and other behaviors in detail, and also label each customer (consumption preferences and usage preferences, etc.) and identify their activity level.

In order to help you better understand the user portrait, I made a picture. Please see the description of the activity level and user tags in the picture below.

(It is particularly important to note that user levels and labels can be freely set in the background according to actual needs; labels and levels are updated once a month, and the system will record the labels and levels obtained by each customer every month.)

In addition to providing user profiling functions, the big data platform also has a user behavior recording function, which records all user usage behaviors from login to leaving. This function allows you to filter out corresponding member data based on relevant search criteria (for example, members who have logged in, shopped, or participated in activities on a certain day).

Some of the filtering conditions are shown below:

(We are in the automotive aftermarket, so we need these screening conditions)

In short, through user portraits and member behavior records based on big data, the problem of who the reminder is sent to can be quickly resolved.

Finally, after the reminder is sent, we need to make a statistics and evaluation of the reminder results, which requires our statistical function to provide data support - because we need to know how many people were reminded under various reminder rules, who were reminded, and what the effect of the reminder was.

Through the above analysis, we already have a very clear idea to carry out specific functional design for this demand. For this, I designed three functions, namely: background reminder statistics, background rule configuration, and client query.

As shown below:

The first is the background reminder statistics, which mainly records how many reminders were issued under each rule and what the specific effects were. Here I divide the reminder records into dormant and non-dormant users, because after a series of reminders in the rules, some users are still inactive and have no consumption behavior. For these customers, I classify them as dormant users and manage them separately.

The following figure shows some of the statistical function prototypes designed:

The second is the background rule configuration, which mainly involves targeted rule configuration for different types of customers. Only by configuring scientific reminder rules according to actual needs and cooperating with the edited reminder templates (SMS templates and telephone reminder voice) can the reminder purpose be better achieved.

Here, based on my own understanding, I have compiled a set of rules, as shown below:

As can be seen from the picture (for reference only), this reminder rule forms a cycle from offline membership conversion to VIP renewal reminders, and then to active reminders for ordinary and VIP members. From the moment a customer registers as an online member, he or she enters into our reminder rules, which means that reminders will accompany the customer throughout the entire life cycle of the product.

It should be noted that this rule only applies to reminders for attracting new customers, promoting activation and conversions, and does not cover reminders for marketing activities . Due to the uncertainty of the time and content of each activity, it is necessary to set scientific reminder rules in advance according to different activity purposes and target customers.

In addition, when we actually compile reminder rules, we should pay special attention to the following issues:

• The reminder frequency should not be too frequent, otherwise it will disgust customers and go against our purpose.
• The reminder content should not be too long, but should be concise, clear and attractive.

The last one is the client query function. This function is mainly based on the rules configured in the background. Operators can filter out user data according to the rules.

It should be noted here that if the reminder platform is connected to the SMS sending platform and the outbound calling platform to realize intelligent reminders, then this is just a simple reminder record query. If the reminder platform is not connected to the SMS platform and the outbound call platform, then this is a data output function. The relevant operation personnel must use this function to filter, summarize and export customer data that meets the rules, and then make non-intelligent reminders through the SMS platform or manual customer service.

In short, the reminder function based on big data and scientific reminder rules will effectively promote new customers, retention, activation and conversion. It is also one of the indispensable tools in the hands of operators. It is also important to remind you that when the number of customers increases, data extraction, analysis and the establishment of user portraits must be planned and developed as a priority, because big data is the foundation of precise operations and the beacon of product planning direction.

Source:lain