The gain brought by the event will be divided into 3 parts: new channel launch + new event fission + old users participating in the event for the first time ws number list To add a new user is to ask for the corresponding new user LTV. It is not listed here for the time being. In addition, why should it be limited to the old users who participate in the activity for the first time? After restricting the old users to participate in the activity for the first time, the daily gain brought by the activity will not be confused with the gain of the old user who has participated in ws number list the activity many times, resulting in the inability to quantify the activity gain well. LT is the activity life cycle duration gain we want to calculate.
LT can be equivalently regarded as the gain of the ws number list daily activity rate of the users in the activity group and the non-participant group in the following N days, and the addition of the daily activity rate gains on the N days is the LT gain. The advantage of choosing to look at the daily activity rate is that we can see the following two points from the trend of the curve, which indirectly verifies the matching ws number list weighting effect of the model. Can ws number list participating and non-participating users be regarded as the same person before participating in the event?
That is, whether the daily activity rate curves of the ws number list two groups of users before participating in the activity overlap, so as to verify the effect of the propensity-weighted score Extending the N-day time period helps us to clearly locate the duration of the activity effect from the change trend of the subsequent retention time ws number list At this point, we have completely completed the gain calculation of the activity effect quantification ROI.