LTV is the sum of the commercial value brought to the enterprise by the loss of user acquisition. It is the ultimate commercial goal of operation improvement and the core indicator of strategy effectiveness verification. Compared with the calculation of the single effect of ROI, LTV is based on the long-term value of user operation. This standard is an analysis Country Email Listand calculation of both weight and quality, which is more conducive to the benign operation of Polaris for long-term acquisition of higher user value.
The applications of LTV are divided into predictive LTV and existing LTV computation. The existing LTV calculation calculates the LTV value of the current user, activity, and channel based on the accumulated data of several months to several years; the LTV prediction algorithm also predicts the LTV value of the future period of time based on the accumulated historical data.
Calculation of LTV: LTV does not have a general fixed formula. Enterprises formulate them according to their own business judgment. The commonly used calculation formulas are LTV=LT*ARPU; LTV=MMR/churn rate.
to make predictions. Usually, there will be a certain range of users according to the accuracy of the data situation. The more data corresponding to the user group, the higher the prediction accuracy.
Application: The application of LTV has applications from planning to analysis and execution. It has two main functions in the planning layer:
Measure user value and operational effectiveness;
Assists in calculating earnings and assisting in the formulation of revenue targets for a future period.
One of the most commonly used analysis models in RFM retail operations, users are divided according to the user's last transaction time, transaction amount, and transaction frequency. The detailed rules and online data are too numerous to go into details. There are two common analysis methods.
User classification through RFM value scoring, used alone or in combination with other tags and data;
Use R/F/M values alone, or combine other labels, data, such as high M val